Re: [FFmpeg-devel] [PATCH] avfilter: add dumpwave filter.

2019-01-08 Thread Tobias Rapp 

On 07.01.2019 15:31, Dmitry Humeniuk wrote:

Signed-off-by: Dmitry Humeniuk 
---
  Changelog|1 +
  doc/filters.texi |   23 +
  libavfilter/Makefile |1 +
  libavfilter/af_dumpwave.c|  234 
  libavfilter/allfilters.c |1 +
  libavfilter/version.h|4 +-
  tests/fate-run.sh|9 +
  tests/fate/filter-audio.mak  |   12 +
  tests/ref/fate/filter-dumpwave   | 1800 ++
  tests/ref/fate/filter-dumpwave-24bit | 1800 ++
  tests/ref/fate/filter-dumpwave-fltp  | 1800 ++
  11 files changed, 5683 insertions(+), 2 deletions(-)
  create mode 100644 libavfilter/af_dumpwave.c
  create mode 100644 tests/ref/fate/filter-dumpwave
  create mode 100644 tests/ref/fate/filter-dumpwave-24bit
  create mode 100644 tests/ref/fate/filter-dumpwave-fltp

diff --git a/Changelog b/Changelog
index f135fa56b6..1eb337796b 100644
--- a/Changelog
+++ b/Changelog
@@ -13,6 +13,7 @@ version :
  - GIF parser
  - vividas demuxer
  - hymt decoder
+- dumpwave filter
  
  
  version 4.1:

diff --git a/doc/filters.texi b/doc/filters.texi
index 98858c5356..66c2961fd8 100644
--- a/doc/filters.texi
+++ b/doc/filters.texi
@@ -2912,6 +2912,29 @@ Set window length in seconds used to split audio into 
segments of equal length.
  Default is 3 seconds.
  @end table
  
+@section dumpwave

+Dump RMS envelope to a file.
+Convert samples to decibels and calculates RMS (Root-Mean-Square) audio power 
in 0 to 1.0 floats.
+
+@table @option
+@item w, width
+Number of data values.
+The default value is @var{1800}
+
+@item n, nb_samples
+Samples count per value per channel, default 128
+
+@item f, file
+Path to a file ``-'' is a shorthand
+for standard output.
+@end table
+
+For example, to generate RMS envelope for 44.1 kHz 6 seconds length audio
+with dimensions @var{1800x140}, samples count @code{44100*6/1800=147} and 
store it to @var{/tmp/out.csv}, you might use:
+@example
+dumpwave=w=1800:n=147:f=/tmp/out.csv
+@end example
+
  @section dynaudnorm
  Dynamic Audio Normalizer.
  
[...]


I appreciate the completeness of the proposed patch (documentation, 
tests, version bump) but it seems that the functionality overlaps with 
existing filters. Besides the different output format I think the same 
can be achieved by using a command like:


ffmpeg -i input.wav -f null -filter:a 
"asetnsamples=147,astats=metadata=on,ametadata=print:key=lavfi.astats.Overall.RMS_peak:file=/tmp/out.dat" 
/dev/null


So if it is only about the output format, it would be better to add CSV 
support to the ametadata filter, IMHO. That way other filters with frame 
metadata output will also gain CSV support.


Best regards,
Tobias

___
ffmpeg-devel mailing list
ffmpeg-devel@ffmpeg.org
http://ffmpeg.org/mailman/listinfo/ffmpeg-devel

Re: [FFmpeg-devel] [PATCH] avfilter: add dumpwave filter.

2018-01-14 Thread Dmytro Humeniuk 
Hi
> On 14 Jan 2018, at 04:04, Michael Niedermayer  wrote:
> 
> On Sat, Jan 13, 2018 at 11:59:59PM +0100, Dmytro Humeniuk wrote:
>> Hi,
>>> On 12 Jan 2018, at 00:20, Dmitry Gumenyuk  wrote:
>>> --
>>> Best regards,
>>> Dmytro
>>> 
 On 11 Jan 2018, at 23:57, Paul B Mahol  wrote:
 
> On 1/11/18, Dmitry Gumenyuk  wrote:
> 
>>> On 11 Jan 2018, at 23:02, Paul B Mahol  wrote:
>>> 
>>> On 1/11/18, Dmitry Gumenyuk  wrote:
>>> 
> On 11 Jan 2018, at 22:41, Paul B Mahol  wrote:
> 
> On 1/11/18, Dmitry Gumenyuk  wrote:
> Hello
> 
> 2018-01-10 11:33 GMT+01:00 Moritz Barsnick :
>>> On Wed, Jan 10, 2018 at 08:58:05 +0100, dmitry.gumen...@gmail.com
>>> wrote:
>>> 
>>> +@table @option
>>> +@item d
>>> +Dimensions @code{WxH}.
>>> +@code{W} - number of data values in json, values will be scaled
>>> according to @code{H}.
>>> +The default value is @var{640x480}
>>> +
>>> +@item s
>>> +Samples count per value per channel
>> 
>> In most other filters/filtersources, s+size is used for dimensions,
>> d+duration for time, and n for an amount/number of frames/samples or
>> so. Might be a good idea to align with this. And use aliases (i.e.
>> implement both "s" and "size").
>> 
>>> +static const AVOption dumpwave_options[] = {
>>> +{ "d", "set width and height", OFFSET(w),
>>> AV_OPT_TYPE_IMAGE_SIZE,
>>> {.str = "640x480"}, 0, 0, FLAGS },
>>> +{ "s", "set number of samples per value per channel",
>>> OFFSET(s),
>>> AV_OPT_TYPE_INT64,  {.i64 = 0}, 0, INT64_MAX, FLAGS },
>>> +{ "json", "set json dump file", OFFSET(json),
>>> AV_OPT_TYPE_STRING,
>>> {
>>> .str = NULL }, 0, 0, FLAGS },
>>> +{ NULL }
>> [...]
>>> +static av_cold int init(AVFilterContext *ctx)
>>> +{
>>> +DumpWaveContext *dumpwave = ctx->priv;
>>> +if(!dumpwave->s) {
>>> +dumpwave->is_disabled = 1;
>>> +av_log(ctx, AV_LOG_ERROR, "Invalid samples per value
>>> config\n");
>>> +}
>>> +return 0;
>> 
>> I don't like the idea of having to provide the "s" parameter. Is
>> there
>> really no useful default?
>> 
>> And now, if s=0, what does the filter do? Just sit around? Couldn't
>> it
>> fail instead?
>> 
>> Apart from that:
>> "if (" is the bracket style ffmpeg uses.
>> 
>>> +if (dumpwave->json && !(dump_fp =
>>> av_fopen_utf8(dumpwave->json,
>>> "w")))
>>> +av_log(ctx, AV_LOG_WARNING, "Flushing dump failed\n");
>> 
>> I would appreciate evaluation of errno and printing the appropriate
>> string (using av_strerror(), I believe).
>> 
>>> +switch (inlink->format) {
>>> +case AV_SAMPLE_FMT_DBLP:
>> 
>> As Kyle hinted: Can this not force a conversion (implicit insertion
>> of
>> aformat filter) to e.g. double by only supporting this format, and
>> reducing this switch to one or two cases? (The filter's output isn't
>> really meant for transparent reuse anyway? af_volumedetect e.g. also
>> supports only one, meaning its output can be a different format than
>> its input.) It's also really hard to go through and later to
>> maintain.
>> It the big switch/case remains, one or two code macros would be
>> appropriate.
> 
> I checked solution used in volumedetect and couldn't find a way to
> read across formats.
 
 I do not understand what you are trying to do.
>>> Sorry, I'm trying to add support for 8, 16, 24, 32, 64 bit sample
>>> formats
> How would you implement such macros? Since version 3.2 astats filter
> uses same approach for reading different formats and as far as I know
> macros harder to debug
 
 astats is using all formats because of numerous reasons. astats uses
 raw
 values,
 your filter just convert each raw value to float representation.
>>> Is this wrong, as I'd like to have high precision?
>> 
>> For rendering to small size image?
> Data can be used for analysis as well. Any size I would say as user may
> define size
 
 Than use floats.
>>> Ok, will give it a try
>> Here is updated patch
 ___
 ffmpeg-devel mailing list
 ffmpeg-devel@ffmpeg.org

Re: [FFmpeg-devel] [PATCH] avfilter: add dumpwave filter.

2018-01-13 Thread Michael Niedermayer 
On Sat, Jan 13, 2018 at 11:59:59PM +0100, Dmytro Humeniuk wrote:
> Hi,
> > On 12 Jan 2018, at 00:20, Dmitry Gumenyuk  wrote:
> > --
> > Best regards,
> > Dmytro
> > 
> >> On 11 Jan 2018, at 23:57, Paul B Mahol  wrote:
> >> 
> >>> On 1/11/18, Dmitry Gumenyuk  wrote:
> >>> 
> > On 11 Jan 2018, at 23:02, Paul B Mahol  wrote:
> > 
> > On 1/11/18, Dmitry Gumenyuk  wrote:
> > 
> >>> On 11 Jan 2018, at 22:41, Paul B Mahol  wrote:
> >>> 
> >>> On 1/11/18, Dmitry Gumenyuk  wrote:
> >>> Hello
> >>> 
> >>> 2018-01-10 11:33 GMT+01:00 Moritz Barsnick :
> > On Wed, Jan 10, 2018 at 08:58:05 +0100, dmitry.gumen...@gmail.com
> > wrote:
> > 
> > +@table @option
> > +@item d
> > +Dimensions @code{WxH}.
> > +@code{W} - number of data values in json, values will be scaled
> > according to @code{H}.
> > +The default value is @var{640x480}
> > +
> > +@item s
> > +Samples count per value per channel
>  
>  In most other filters/filtersources, s+size is used for dimensions,
>  d+duration for time, and n for an amount/number of frames/samples or
>  so. Might be a good idea to align with this. And use aliases (i.e.
>  implement both "s" and "size").
>  
> > +static const AVOption dumpwave_options[] = {
> > +{ "d", "set width and height", OFFSET(w),
> > AV_OPT_TYPE_IMAGE_SIZE,
> > {.str = "640x480"}, 0, 0, FLAGS },
> > +{ "s", "set number of samples per value per channel",
> > OFFSET(s),
> > AV_OPT_TYPE_INT64,  {.i64 = 0}, 0, INT64_MAX, FLAGS },
> > +{ "json", "set json dump file", OFFSET(json),
> > AV_OPT_TYPE_STRING,
> > {
> > .str = NULL }, 0, 0, FLAGS },
> > +{ NULL }
>  [...]
> > +static av_cold int init(AVFilterContext *ctx)
> > +{
> > +DumpWaveContext *dumpwave = ctx->priv;
> > +if(!dumpwave->s) {
> > +dumpwave->is_disabled = 1;
> > +av_log(ctx, AV_LOG_ERROR, "Invalid samples per value
> > config\n");
> > +}
> > +return 0;
>  
>  I don't like the idea of having to provide the "s" parameter. Is
>  there
>  really no useful default?
>  
>  And now, if s=0, what does the filter do? Just sit around? Couldn't
>  it
>  fail instead?
>  
>  Apart from that:
>  "if (" is the bracket style ffmpeg uses.
>  
> > +if (dumpwave->json && !(dump_fp =
> > av_fopen_utf8(dumpwave->json,
> > "w")))
> > +av_log(ctx, AV_LOG_WARNING, "Flushing dump failed\n");
>  
>  I would appreciate evaluation of errno and printing the appropriate
>  string (using av_strerror(), I believe).
>  
> > +switch (inlink->format) {
> > +case AV_SAMPLE_FMT_DBLP:
>  
>  As Kyle hinted: Can this not force a conversion (implicit insertion
>  of
>  aformat filter) to e.g. double by only supporting this format, and
>  reducing this switch to one or two cases? (The filter's output isn't
>  really meant for transparent reuse anyway? af_volumedetect e.g. also
>  supports only one, meaning its output can be a different format than
>  its input.) It's also really hard to go through and later to
>  maintain.
>  It the big switch/case remains, one or two code macros would be
>  appropriate.
> >>> 
> >>> I checked solution used in volumedetect and couldn't find a way to
> >>> read across formats.
> >> 
> >> I do not understand what you are trying to do.
> > Sorry, I'm trying to add support for 8, 16, 24, 32, 64 bit sample
> > formats
> >>> How would you implement such macros? Since version 3.2 astats filter
> >>> uses same approach for reading different formats and as far as I know
> >>> macros harder to debug
> >> 
> >> astats is using all formats because of numerous reasons. astats uses
> >> raw
> >> values,
> >> your filter just convert each raw value to float representation.
> > Is this wrong, as I'd like to have high precision?
>  
>  For rendering to small size image?
> >>> Data can be used for analysis as well. Any size I would say as user may
> >>> define size
> >> 
> >> Than use floats.
> > Ok, will give it a try
> Here is updated patch
> >> ___
> >> ffmpeg-devel mailing list
> >> ffmpeg-devel@ffmpeg.org
> >> http://ffmpeg.org/mailman/listinfo/ffmpeg-devel

>  Changelog

Re: [FFmpeg-devel] [PATCH] avfilter: add dumpwave filter.

2018-01-13 Thread Dmytro Humeniuk 
Hi,
> On 12 Jan 2018, at 00:20, Dmitry Gumenyuk  wrote:
> --
> Best regards,
> Dmytro
> 
>> On 11 Jan 2018, at 23:57, Paul B Mahol  wrote:
>> 
>>> On 1/11/18, Dmitry Gumenyuk  wrote:
>>> 
> On 11 Jan 2018, at 23:02, Paul B Mahol  wrote:
> 
> On 1/11/18, Dmitry Gumenyuk  wrote:
> 
>>> On 11 Jan 2018, at 22:41, Paul B Mahol  wrote:
>>> 
>>> On 1/11/18, Dmitry Gumenyuk  wrote:
>>> Hello
>>> 
>>> 2018-01-10 11:33 GMT+01:00 Moritz Barsnick :
> On Wed, Jan 10, 2018 at 08:58:05 +0100, dmitry.gumen...@gmail.com
> wrote:
> 
> +@table @option
> +@item d
> +Dimensions @code{WxH}.
> +@code{W} - number of data values in json, values will be scaled
> according to @code{H}.
> +The default value is @var{640x480}
> +
> +@item s
> +Samples count per value per channel
 
 In most other filters/filtersources, s+size is used for dimensions,
 d+duration for time, and n for an amount/number of frames/samples or
 so. Might be a good idea to align with this. And use aliases (i.e.
 implement both "s" and "size").
 
> +static const AVOption dumpwave_options[] = {
> +{ "d", "set width and height", OFFSET(w),
> AV_OPT_TYPE_IMAGE_SIZE,
> {.str = "640x480"}, 0, 0, FLAGS },
> +{ "s", "set number of samples per value per channel",
> OFFSET(s),
> AV_OPT_TYPE_INT64,  {.i64 = 0}, 0, INT64_MAX, FLAGS },
> +{ "json", "set json dump file", OFFSET(json),
> AV_OPT_TYPE_STRING,
> {
> .str = NULL }, 0, 0, FLAGS },
> +{ NULL }
 [...]
> +static av_cold int init(AVFilterContext *ctx)
> +{
> +DumpWaveContext *dumpwave = ctx->priv;
> +if(!dumpwave->s) {
> +dumpwave->is_disabled = 1;
> +av_log(ctx, AV_LOG_ERROR, "Invalid samples per value
> config\n");
> +}
> +return 0;
 
 I don't like the idea of having to provide the "s" parameter. Is
 there
 really no useful default?
 
 And now, if s=0, what does the filter do? Just sit around? Couldn't
 it
 fail instead?
 
 Apart from that:
 "if (" is the bracket style ffmpeg uses.
 
> +if (dumpwave->json && !(dump_fp =
> av_fopen_utf8(dumpwave->json,
> "w")))
> +av_log(ctx, AV_LOG_WARNING, "Flushing dump failed\n");
 
 I would appreciate evaluation of errno and printing the appropriate
 string (using av_strerror(), I believe).
 
> +switch (inlink->format) {
> +case AV_SAMPLE_FMT_DBLP:
 
 As Kyle hinted: Can this not force a conversion (implicit insertion
 of
 aformat filter) to e.g. double by only supporting this format, and
 reducing this switch to one or two cases? (The filter's output isn't
 really meant for transparent reuse anyway? af_volumedetect e.g. also
 supports only one, meaning its output can be a different format than
 its input.) It's also really hard to go through and later to
 maintain.
 It the big switch/case remains, one or two code macros would be
 appropriate.
>>> 
>>> I checked solution used in volumedetect and couldn't find a way to
>>> read across formats.
>> 
>> I do not understand what you are trying to do.
> Sorry, I'm trying to add support for 8, 16, 24, 32, 64 bit sample
> formats
>>> How would you implement such macros? Since version 3.2 astats filter
>>> uses same approach for reading different formats and as far as I know
>>> macros harder to debug
>> 
>> astats is using all formats because of numerous reasons. astats uses
>> raw
>> values,
>> your filter just convert each raw value to float representation.
> Is this wrong, as I'd like to have high precision?
 
 For rendering to small size image?
>>> Data can be used for analysis as well. Any size I would say as user may
>>> define size
>> 
>> Than use floats.
> Ok, will give it a try
Here is updated patch
>> ___
>> ffmpeg-devel mailing list
>> ffmpeg-devel@ffmpeg.org
>> http://ffmpeg.org/mailman/listinfo/ffmpeg-devel


FFmpeg-devel-v2-avfilter-add-dumpwave-filter.patch
Description: Binary data


___
ffmpeg-devel mailing list
ffmpeg-devel@ffmpeg.org
http://ffmpeg.org/mailman/listinfo/ffmpeg-devel

Re: [FFmpeg-devel] [PATCH] avfilter: add dumpwave filter.

2018-01-11 Thread Dmitry Gumenyuk 


--
Best regards,
Dmytro

> On 11 Jan 2018, at 23:57, Paul B Mahol  wrote:
> 
>> On 1/11/18, Dmitry Gumenyuk  wrote:
>> 
 On 11 Jan 2018, at 23:02, Paul B Mahol  wrote:
 
 On 1/11/18, Dmitry Gumenyuk  wrote:
 
>> On 11 Jan 2018, at 22:41, Paul B Mahol  wrote:
>> 
>> On 1/11/18, Dmitry Gumenyuk  wrote:
>> Hello
>> 
>> 2018-01-10 11:33 GMT+01:00 Moritz Barsnick :
 On Wed, Jan 10, 2018 at 08:58:05 +0100, dmitry.gumen...@gmail.com
 wrote:
 
 +@table @option
 +@item d
 +Dimensions @code{WxH}.
 +@code{W} - number of data values in json, values will be scaled
 according to @code{H}.
 +The default value is @var{640x480}
 +
 +@item s
 +Samples count per value per channel
>>> 
>>> In most other filters/filtersources, s+size is used for dimensions,
>>> d+duration for time, and n for an amount/number of frames/samples or
>>> so. Might be a good idea to align with this. And use aliases (i.e.
>>> implement both "s" and "size").
>>> 
 +static const AVOption dumpwave_options[] = {
 +{ "d", "set width and height", OFFSET(w),
 AV_OPT_TYPE_IMAGE_SIZE,
 {.str = "640x480"}, 0, 0, FLAGS },
 +{ "s", "set number of samples per value per channel",
 OFFSET(s),
 AV_OPT_TYPE_INT64,  {.i64 = 0}, 0, INT64_MAX, FLAGS },
 +{ "json", "set json dump file", OFFSET(json),
 AV_OPT_TYPE_STRING,
 {
 .str = NULL }, 0, 0, FLAGS },
 +{ NULL }
>>> [...]
 +static av_cold int init(AVFilterContext *ctx)
 +{
 +DumpWaveContext *dumpwave = ctx->priv;
 +if(!dumpwave->s) {
 +dumpwave->is_disabled = 1;
 +av_log(ctx, AV_LOG_ERROR, "Invalid samples per value
 config\n");
 +}
 +return 0;
>>> 
>>> I don't like the idea of having to provide the "s" parameter. Is
>>> there
>>> really no useful default?
>>> 
>>> And now, if s=0, what does the filter do? Just sit around? Couldn't
>>> it
>>> fail instead?
>>> 
>>> Apart from that:
>>> "if (" is the bracket style ffmpeg uses.
>>> 
 +if (dumpwave->json && !(dump_fp =
 av_fopen_utf8(dumpwave->json,
 "w")))
 +av_log(ctx, AV_LOG_WARNING, "Flushing dump failed\n");
>>> 
>>> I would appreciate evaluation of errno and printing the appropriate
>>> string (using av_strerror(), I believe).
>>> 
 +switch (inlink->format) {
 +case AV_SAMPLE_FMT_DBLP:
>>> 
>>> As Kyle hinted: Can this not force a conversion (implicit insertion
>>> of
>>> aformat filter) to e.g. double by only supporting this format, and
>>> reducing this switch to one or two cases? (The filter's output isn't
>>> really meant for transparent reuse anyway? af_volumedetect e.g. also
>>> supports only one, meaning its output can be a different format than
>>> its input.) It's also really hard to go through and later to
>>> maintain.
>>> It the big switch/case remains, one or two code macros would be
>>> appropriate.
>> 
>> I checked solution used in volumedetect and couldn't find a way to
>> read across formats.
> 
> I do not understand what you are trying to do.
 Sorry, I'm trying to add support for 8, 16, 24, 32, 64 bit sample
 formats
>> How would you implement such macros? Since version 3.2 astats filter
>> uses same approach for reading different formats and as far as I know
>> macros harder to debug
> 
> astats is using all formats because of numerous reasons. astats uses
> raw
> values,
> your filter just convert each raw value to float representation.
 Is this wrong, as I'd like to have high precision?
>>> 
>>> For rendering to small size image?
>> Data can be used for analysis as well. Any size I would say as user may
>> define size
> 
> Than use floats.
Ok, will give it a try
> ___
> ffmpeg-devel mailing list
> ffmpeg-devel@ffmpeg.org
> http://ffmpeg.org/mailman/listinfo/ffmpeg-devel


smime.p7s
Description: S/MIME cryptographic signature
___
ffmpeg-devel mailing list
ffmpeg-devel@ffmpeg.org
http://ffmpeg.org/mailman/listinfo/ffmpeg-devel

Re: [FFmpeg-devel] [PATCH] avfilter: add dumpwave filter.

2018-01-11 Thread Paul B Mahol 
On 1/11/18, Dmitry Gumenyuk  wrote:
>
>> On 11 Jan 2018, at 23:02, Paul B Mahol  wrote:
>>
>>> On 1/11/18, Dmitry Gumenyuk  wrote:
>>>
> On 11 Jan 2018, at 22:41, Paul B Mahol  wrote:
>
> On 1/11/18, Dmitry Gumenyuk  wrote:
> Hello
>
> 2018-01-10 11:33 GMT+01:00 Moritz Barsnick :
>>> On Wed, Jan 10, 2018 at 08:58:05 +0100, dmitry.gumen...@gmail.com
>>> wrote:
>>>
>>> +@table @option
>>> +@item d
>>> +Dimensions @code{WxH}.
>>> +@code{W} - number of data values in json, values will be scaled
>>> according to @code{H}.
>>> +The default value is @var{640x480}
>>> +
>>> +@item s
>>> +Samples count per value per channel
>>
>> In most other filters/filtersources, s+size is used for dimensions,
>> d+duration for time, and n for an amount/number of frames/samples or
>> so. Might be a good idea to align with this. And use aliases (i.e.
>> implement both "s" and "size").
>>
>>> +static const AVOption dumpwave_options[] = {
>>> +{ "d", "set width and height", OFFSET(w),
>>> AV_OPT_TYPE_IMAGE_SIZE,
>>> {.str = "640x480"}, 0, 0, FLAGS },
>>> +{ "s", "set number of samples per value per channel",
>>> OFFSET(s),
>>> AV_OPT_TYPE_INT64,  {.i64 = 0}, 0, INT64_MAX, FLAGS },
>>> +{ "json", "set json dump file", OFFSET(json),
>>> AV_OPT_TYPE_STRING,
>>> {
>>> .str = NULL }, 0, 0, FLAGS },
>>> +{ NULL }
>> [...]
>>> +static av_cold int init(AVFilterContext *ctx)
>>> +{
>>> +DumpWaveContext *dumpwave = ctx->priv;
>>> +if(!dumpwave->s) {
>>> +dumpwave->is_disabled = 1;
>>> +av_log(ctx, AV_LOG_ERROR, "Invalid samples per value
>>> config\n");
>>> +}
>>> +return 0;
>>
>> I don't like the idea of having to provide the "s" parameter. Is
>> there
>> really no useful default?
>>
>> And now, if s=0, what does the filter do? Just sit around? Couldn't
>> it
>> fail instead?
>>
>> Apart from that:
>> "if (" is the bracket style ffmpeg uses.
>>
>>> +if (dumpwave->json && !(dump_fp =
>>> av_fopen_utf8(dumpwave->json,
>>> "w")))
>>> +av_log(ctx, AV_LOG_WARNING, "Flushing dump failed\n");
>>
>> I would appreciate evaluation of errno and printing the appropriate
>> string (using av_strerror(), I believe).
>>
>>> +switch (inlink->format) {
>>> +case AV_SAMPLE_FMT_DBLP:
>>
>> As Kyle hinted: Can this not force a conversion (implicit insertion
>> of
>> aformat filter) to e.g. double by only supporting this format, and
>> reducing this switch to one or two cases? (The filter's output isn't
>> really meant for transparent reuse anyway? af_volumedetect e.g. also
>> supports only one, meaning its output can be a different format than
>> its input.) It's also really hard to go through and later to
>> maintain.
>> It the big switch/case remains, one or two code macros would be
>> appropriate.
>
> I checked solution used in volumedetect and couldn't find a way to
> read across formats.

 I do not understand what you are trying to do.
>>> Sorry, I'm trying to add support for 8, 16, 24, 32, 64 bit sample
>>> formats
> How would you implement such macros? Since version 3.2 astats filter
> uses same approach for reading different formats and as far as I know
> macros harder to debug

 astats is using all formats because of numerous reasons. astats uses
 raw
 values,
 your filter just convert each raw value to float representation.
>>> Is this wrong, as I'd like to have high precision?
>>
>> For rendering to small size image?
> Data can be used for analysis as well. Any size I would say as user may
> define size

Than use floats.
___
ffmpeg-devel mailing list
ffmpeg-devel@ffmpeg.org
http://ffmpeg.org/mailman/listinfo/ffmpeg-devel

Re: [FFmpeg-devel] [PATCH] avfilter: add dumpwave filter.

2018-01-11 Thread Dmitry Gumenyuk 

> On 11 Jan 2018, at 23:02, Paul B Mahol  wrote:
> 
>> On 1/11/18, Dmitry Gumenyuk  wrote:
>> 
 On 11 Jan 2018, at 22:41, Paul B Mahol  wrote:
 
 On 1/11/18, Dmitry Gumenyuk  wrote:
 Hello
 
 2018-01-10 11:33 GMT+01:00 Moritz Barsnick :
>> On Wed, Jan 10, 2018 at 08:58:05 +0100, dmitry.gumen...@gmail.com
>> wrote:
>> 
>> +@table @option
>> +@item d
>> +Dimensions @code{WxH}.
>> +@code{W} - number of data values in json, values will be scaled
>> according to @code{H}.
>> +The default value is @var{640x480}
>> +
>> +@item s
>> +Samples count per value per channel
> 
> In most other filters/filtersources, s+size is used for dimensions,
> d+duration for time, and n for an amount/number of frames/samples or
> so. Might be a good idea to align with this. And use aliases (i.e.
> implement both "s" and "size").
> 
>> +static const AVOption dumpwave_options[] = {
>> +{ "d", "set width and height", OFFSET(w), AV_OPT_TYPE_IMAGE_SIZE,
>> {.str = "640x480"}, 0, 0, FLAGS },
>> +{ "s", "set number of samples per value per channel",  OFFSET(s),
>> AV_OPT_TYPE_INT64,  {.i64 = 0}, 0, INT64_MAX, FLAGS },
>> +{ "json", "set json dump file", OFFSET(json), AV_OPT_TYPE_STRING,
>> {
>> .str = NULL }, 0, 0, FLAGS },
>> +{ NULL }
> [...]
>> +static av_cold int init(AVFilterContext *ctx)
>> +{
>> +DumpWaveContext *dumpwave = ctx->priv;
>> +if(!dumpwave->s) {
>> +dumpwave->is_disabled = 1;
>> +av_log(ctx, AV_LOG_ERROR, "Invalid samples per value
>> config\n");
>> +}
>> +return 0;
> 
> I don't like the idea of having to provide the "s" parameter. Is there
> really no useful default?
> 
> And now, if s=0, what does the filter do? Just sit around? Couldn't it
> fail instead?
> 
> Apart from that:
> "if (" is the bracket style ffmpeg uses.
> 
>> +if (dumpwave->json && !(dump_fp =
>> av_fopen_utf8(dumpwave->json,
>> "w")))
>> +av_log(ctx, AV_LOG_WARNING, "Flushing dump failed\n");
> 
> I would appreciate evaluation of errno and printing the appropriate
> string (using av_strerror(), I believe).
> 
>> +switch (inlink->format) {
>> +case AV_SAMPLE_FMT_DBLP:
> 
> As Kyle hinted: Can this not force a conversion (implicit insertion of
> aformat filter) to e.g. double by only supporting this format, and
> reducing this switch to one or two cases? (The filter's output isn't
> really meant for transparent reuse anyway? af_volumedetect e.g. also
> supports only one, meaning its output can be a different format than
> its input.) It's also really hard to go through and later to maintain.
> It the big switch/case remains, one or two code macros would be
> appropriate.
 
 I checked solution used in volumedetect and couldn't find a way to
 read across formats.
>>> 
>>> I do not understand what you are trying to do.
>> Sorry, I'm trying to add support for 8, 16, 24, 32, 64 bit sample formats
 How would you implement such macros? Since version 3.2 astats filter
 uses same approach for reading different formats and as far as I know
 macros harder to debug
>>> 
>>> astats is using all formats because of numerous reasons. astats uses raw
>>> values,
>>> your filter just convert each raw value to float representation.
>> Is this wrong, as I'd like to have high precision?
> 
> For rendering to small size image?
Data can be used for analysis as well. Any size I would say as user may define 
size
> ___
> ffmpeg-devel mailing list
> ffmpeg-devel@ffmpeg.org
> http://ffmpeg.org/mailman/listinfo/ffmpeg-devel


smime.p7s
Description: S/MIME cryptographic signature
___
ffmpeg-devel mailing list
ffmpeg-devel@ffmpeg.org
http://ffmpeg.org/mailman/listinfo/ffmpeg-devel

Re: [FFmpeg-devel] [PATCH] avfilter: add dumpwave filter.

2018-01-11 Thread Dmitry Gumenyuk 

> On 11 Jan 2018, at 22:41, Paul B Mahol  wrote:
> 
>> On 1/11/18, Dmitry Gumenyuk  wrote:
>> Hello
>> 
>> 2018-01-10 11:33 GMT+01:00 Moritz Barsnick :
 On Wed, Jan 10, 2018 at 08:58:05 +0100, dmitry.gumen...@gmail.com wrote:
 
 +@table @option
 +@item d
 +Dimensions @code{WxH}.
 +@code{W} - number of data values in json, values will be scaled
 according to @code{H}.
 +The default value is @var{640x480}
 +
 +@item s
 +Samples count per value per channel
>>> 
>>> In most other filters/filtersources, s+size is used for dimensions,
>>> d+duration for time, and n for an amount/number of frames/samples or
>>> so. Might be a good idea to align with this. And use aliases (i.e.
>>> implement both "s" and "size").
>>> 
 +static const AVOption dumpwave_options[] = {
 +{ "d", "set width and height", OFFSET(w), AV_OPT_TYPE_IMAGE_SIZE,
 {.str = "640x480"}, 0, 0, FLAGS },
 +{ "s", "set number of samples per value per channel",  OFFSET(s),
 AV_OPT_TYPE_INT64,  {.i64 = 0}, 0, INT64_MAX, FLAGS },
 +{ "json", "set json dump file", OFFSET(json), AV_OPT_TYPE_STRING, {
 .str = NULL }, 0, 0, FLAGS },
 +{ NULL }
>>> [...]
 +static av_cold int init(AVFilterContext *ctx)
 +{
 +DumpWaveContext *dumpwave = ctx->priv;
 +if(!dumpwave->s) {
 +dumpwave->is_disabled = 1;
 +av_log(ctx, AV_LOG_ERROR, "Invalid samples per value config\n");
 +}
 +return 0;
>>> 
>>> I don't like the idea of having to provide the "s" parameter. Is there
>>> really no useful default?
>>> 
>>> And now, if s=0, what does the filter do? Just sit around? Couldn't it
>>> fail instead?
>>> 
>>> Apart from that:
>>> "if (" is the bracket style ffmpeg uses.
>>> 
 +if (dumpwave->json && !(dump_fp = av_fopen_utf8(dumpwave->json,
 "w")))
 +av_log(ctx, AV_LOG_WARNING, "Flushing dump failed\n");
>>> 
>>> I would appreciate evaluation of errno and printing the appropriate
>>> string (using av_strerror(), I believe).
>>> 
 +switch (inlink->format) {
 +case AV_SAMPLE_FMT_DBLP:
>>> 
>>> As Kyle hinted: Can this not force a conversion (implicit insertion of
>>> aformat filter) to e.g. double by only supporting this format, and
>>> reducing this switch to one or two cases? (The filter's output isn't
>>> really meant for transparent reuse anyway? af_volumedetect e.g. also
>>> supports only one, meaning its output can be a different format than
>>> its input.) It's also really hard to go through and later to maintain.
>>> It the big switch/case remains, one or two code macros would be
>>> appropriate.
>> 
>> I checked solution used in volumedetect and couldn't find a way to
>> read across formats.
> 
> I do not understand what you are trying to do.
Sorry, I'm trying to add support for 8, 16, 24, 32, 64 bit sample formats
>> How would you implement such macros? Since version 3.2 astats filter
>> uses same approach for reading different formats and as far as I know
>> macros harder to debug
> 
> astats is using all formats because of numerous reasons. astats uses raw 
> values,
> your filter just convert each raw value to float representation.
Is this wrong, as I'd like to have high precision?
> This filter does not need support for so many formats, it could
> support just s16.
> ___
> ffmpeg-devel mailing list
> ffmpeg-devel@ffmpeg.org
> http://ffmpeg.org/mailman/listinfo/ffmpeg-devel


smime.p7s
Description: S/MIME cryptographic signature
___
ffmpeg-devel mailing list
ffmpeg-devel@ffmpeg.org
http://ffmpeg.org/mailman/listinfo/ffmpeg-devel

Re: [FFmpeg-devel] [PATCH] avfilter: add dumpwave filter.

2018-01-11 Thread Paul B Mahol 
On 1/11/18, Dmitry Gumenyuk  wrote:
>
>> On 11 Jan 2018, at 22:41, Paul B Mahol  wrote:
>>
>>> On 1/11/18, Dmitry Gumenyuk  wrote:
>>> Hello
>>>
>>> 2018-01-10 11:33 GMT+01:00 Moritz Barsnick :
> On Wed, Jan 10, 2018 at 08:58:05 +0100, dmitry.gumen...@gmail.com
> wrote:
>
> +@table @option
> +@item d
> +Dimensions @code{WxH}.
> +@code{W} - number of data values in json, values will be scaled
> according to @code{H}.
> +The default value is @var{640x480}
> +
> +@item s
> +Samples count per value per channel

 In most other filters/filtersources, s+size is used for dimensions,
 d+duration for time, and n for an amount/number of frames/samples or
 so. Might be a good idea to align with this. And use aliases (i.e.
 implement both "s" and "size").

> +static const AVOption dumpwave_options[] = {
> +{ "d", "set width and height", OFFSET(w), AV_OPT_TYPE_IMAGE_SIZE,
> {.str = "640x480"}, 0, 0, FLAGS },
> +{ "s", "set number of samples per value per channel",  OFFSET(s),
> AV_OPT_TYPE_INT64,  {.i64 = 0}, 0, INT64_MAX, FLAGS },
> +{ "json", "set json dump file", OFFSET(json), AV_OPT_TYPE_STRING,
> {
> .str = NULL }, 0, 0, FLAGS },
> +{ NULL }
 [...]
> +static av_cold int init(AVFilterContext *ctx)
> +{
> +DumpWaveContext *dumpwave = ctx->priv;
> +if(!dumpwave->s) {
> +dumpwave->is_disabled = 1;
> +av_log(ctx, AV_LOG_ERROR, "Invalid samples per value
> config\n");
> +}
> +return 0;

 I don't like the idea of having to provide the "s" parameter. Is there
 really no useful default?

 And now, if s=0, what does the filter do? Just sit around? Couldn't it
 fail instead?

 Apart from that:
 "if (" is the bracket style ffmpeg uses.

> +if (dumpwave->json && !(dump_fp =
> av_fopen_utf8(dumpwave->json,
> "w")))
> +av_log(ctx, AV_LOG_WARNING, "Flushing dump failed\n");

 I would appreciate evaluation of errno and printing the appropriate
 string (using av_strerror(), I believe).

> +switch (inlink->format) {
> +case AV_SAMPLE_FMT_DBLP:

 As Kyle hinted: Can this not force a conversion (implicit insertion of
 aformat filter) to e.g. double by only supporting this format, and
 reducing this switch to one or two cases? (The filter's output isn't
 really meant for transparent reuse anyway? af_volumedetect e.g. also
 supports only one, meaning its output can be a different format than
 its input.) It's also really hard to go through and later to maintain.
 It the big switch/case remains, one or two code macros would be
 appropriate.
>>>
>>> I checked solution used in volumedetect and couldn't find a way to
>>> read across formats.
>>
>> I do not understand what you are trying to do.
> Sorry, I'm trying to add support for 8, 16, 24, 32, 64 bit sample formats
>>> How would you implement such macros? Since version 3.2 astats filter
>>> uses same approach for reading different formats and as far as I know
>>> macros harder to debug
>>
>> astats is using all formats because of numerous reasons. astats uses raw
>> values,
>> your filter just convert each raw value to float representation.
> Is this wrong, as I'd like to have high precision?

For rendering to small size image?
___
ffmpeg-devel mailing list
ffmpeg-devel@ffmpeg.org
http://ffmpeg.org/mailman/listinfo/ffmpeg-devel

Re: [FFmpeg-devel] [PATCH] avfilter: add dumpwave filter.

2018-01-11 Thread Paul B Mahol 
On 1/11/18, Dmitry Gumenyuk  wrote:
> Hello
>
> 2018-01-10 11:33 GMT+01:00 Moritz Barsnick :
>> On Wed, Jan 10, 2018 at 08:58:05 +0100, dmitry.gumen...@gmail.com wrote:
>>
>>> +@table @option
>>> +@item d
>>> +Dimensions @code{WxH}.
>>> +@code{W} - number of data values in json, values will be scaled
>>> according to @code{H}.
>>> +The default value is @var{640x480}
>>> +
>>> +@item s
>>> +Samples count per value per channel
>>
>> In most other filters/filtersources, s+size is used for dimensions,
>> d+duration for time, and n for an amount/number of frames/samples or
>> so. Might be a good idea to align with this. And use aliases (i.e.
>> implement both "s" and "size").
>>
>>> +static const AVOption dumpwave_options[] = {
>>> +{ "d", "set width and height", OFFSET(w), AV_OPT_TYPE_IMAGE_SIZE,
>>> {.str = "640x480"}, 0, 0, FLAGS },
>>> +{ "s", "set number of samples per value per channel",  OFFSET(s),
>>> AV_OPT_TYPE_INT64,  {.i64 = 0}, 0, INT64_MAX, FLAGS },
>>> +{ "json", "set json dump file", OFFSET(json), AV_OPT_TYPE_STRING, {
>>> .str = NULL }, 0, 0, FLAGS },
>>> +{ NULL }
>> [...]
>>> +static av_cold int init(AVFilterContext *ctx)
>>> +{
>>> +DumpWaveContext *dumpwave = ctx->priv;
>>> +if(!dumpwave->s) {
>>> +dumpwave->is_disabled = 1;
>>> +av_log(ctx, AV_LOG_ERROR, "Invalid samples per value config\n");
>>> +}
>>> +return 0;
>>
>> I don't like the idea of having to provide the "s" parameter. Is there
>> really no useful default?
>>
>> And now, if s=0, what does the filter do? Just sit around? Couldn't it
>> fail instead?
>>
>> Apart from that:
>> "if (" is the bracket style ffmpeg uses.
>>
>>> +if (dumpwave->json && !(dump_fp = av_fopen_utf8(dumpwave->json,
>>> "w")))
>>> +av_log(ctx, AV_LOG_WARNING, "Flushing dump failed\n");
>>
>> I would appreciate evaluation of errno and printing the appropriate
>> string (using av_strerror(), I believe).
>>
>>> +switch (inlink->format) {
>>> +case AV_SAMPLE_FMT_DBLP:
>>
>> As Kyle hinted: Can this not force a conversion (implicit insertion of
>> aformat filter) to e.g. double by only supporting this format, and
>> reducing this switch to one or two cases? (The filter's output isn't
>> really meant for transparent reuse anyway? af_volumedetect e.g. also
>> supports only one, meaning its output can be a different format than
>> its input.) It's also really hard to go through and later to maintain.
>> It the big switch/case remains, one or two code macros would be
>> appropriate.
>
> I checked solution used in volumedetect and couldn't find a way to
> read across formats.

I do not understand what you are trying to do.

> How would you implement such macros? Since version 3.2 astats filter
> uses same approach for reading different formats and as far as I know
> macros harder to debug

astats is using all formats because of numerous reasons. astats uses raw values,
your filter just convert each raw value to float representation.

This filter does not need support for so many formats, it could
support just s16.
___
ffmpeg-devel mailing list
ffmpeg-devel@ffmpeg.org
http://ffmpeg.org/mailman/listinfo/ffmpeg-devel

Re: [FFmpeg-devel] [PATCH] avfilter: add dumpwave filter.

2018-01-11 Thread Dmitry Gumenyuk 
Hello

2018-01-10 11:33 GMT+01:00 Moritz Barsnick :
> On Wed, Jan 10, 2018 at 08:58:05 +0100, dmitry.gumen...@gmail.com wrote:
>
>> +@table @option
>> +@item d
>> +Dimensions @code{WxH}.
>> +@code{W} - number of data values in json, values will be scaled according 
>> to @code{H}.
>> +The default value is @var{640x480}
>> +
>> +@item s
>> +Samples count per value per channel
>
> In most other filters/filtersources, s+size is used for dimensions,
> d+duration for time, and n for an amount/number of frames/samples or
> so. Might be a good idea to align with this. And use aliases (i.e.
> implement both "s" and "size").
>
>> +static const AVOption dumpwave_options[] = {
>> +{ "d", "set width and height", OFFSET(w), AV_OPT_TYPE_IMAGE_SIZE, {.str 
>> = "640x480"}, 0, 0, FLAGS },
>> +{ "s", "set number of samples per value per channel",  OFFSET(s), 
>> AV_OPT_TYPE_INT64,  {.i64 = 0}, 0, INT64_MAX, FLAGS },
>> +{ "json", "set json dump file", OFFSET(json), AV_OPT_TYPE_STRING, { 
>> .str = NULL }, 0, 0, FLAGS },
>> +{ NULL }
> [...]
>> +static av_cold int init(AVFilterContext *ctx)
>> +{
>> +DumpWaveContext *dumpwave = ctx->priv;
>> +if(!dumpwave->s) {
>> +dumpwave->is_disabled = 1;
>> +av_log(ctx, AV_LOG_ERROR, "Invalid samples per value config\n");
>> +}
>> +return 0;
>
> I don't like the idea of having to provide the "s" parameter. Is there
> really no useful default?
>
> And now, if s=0, what does the filter do? Just sit around? Couldn't it
> fail instead?
>
> Apart from that:
> "if (" is the bracket style ffmpeg uses.
>
>> +if (dumpwave->json && !(dump_fp = av_fopen_utf8(dumpwave->json, 
>> "w")))
>> +av_log(ctx, AV_LOG_WARNING, "Flushing dump failed\n");
>
> I would appreciate evaluation of errno and printing the appropriate
> string (using av_strerror(), I believe).
>
>> +switch (inlink->format) {
>> +case AV_SAMPLE_FMT_DBLP:
>
> As Kyle hinted: Can this not force a conversion (implicit insertion of
> aformat filter) to e.g. double by only supporting this format, and
> reducing this switch to one or two cases? (The filter's output isn't
> really meant for transparent reuse anyway? af_volumedetect e.g. also
> supports only one, meaning its output can be a different format than
> its input.) It's also really hard to go through and later to maintain.
> It the big switch/case remains, one or two code macros would be
> appropriate.

I checked solution used in volumedetect and couldn't find a way to
read across formats.
How would you implement such macros? Since version 3.2 astats filter
uses same approach for reading different formats and as far as I know
macros harder to debug

>> +for (i = 0; i < buf->nb_samples; i++) {
>> +for (c = 0; c < channels; c++, src++)
>> +calc_db_rms(dumpwave, *src);
>> +}}
>
> I believe the curly brackets need to be placed differently.
>
>> +.description   = NULL_IF_CONFIG_SMALL("Dumps RMS amplitude to JSON 
>> file"),
>
> Imperative, i.e. "Dump RMS ...".
>
>> +FATE_AFILTER-$(call FILTERDEMDEC, DUMPWAVE, WAV, PCM_S16LE) += 
>> fate-filter-dumpwave
>> +fate-filter-dumpwave: SRC = $(TARGET_PATH)/tests/data/asynth-44100-2.wav
>> +fate-filter-dumpwave: CMD = ffmpeg -i $(SRC) -af 
>> dumpwave=d=1800x140:s=147:json=$(TARGET_PATH)/tests/data/fate/filter-dumpwave.out
>>  -f null - && cat $(TARGET_PATH)/tests/data/fate/filter-dumpwave.out
>> +
>
> If you need to implement all formats, you might want to test them all.
> Just a thought.
>
> Moritz
> ___
> ffmpeg-devel mailing list
> ffmpeg-devel@ffmpeg.org
> http://ffmpeg.org/mailman/listinfo/ffmpeg-devel
___
ffmpeg-devel mailing list
ffmpeg-devel@ffmpeg.org
http://ffmpeg.org/mailman/listinfo/ffmpeg-devel

Re: [FFmpeg-devel] [PATCH] avfilter: add dumpwave filter.

2018-01-10 Thread Moritz Barsnick 
On Wed, Jan 10, 2018 at 08:58:05 +0100, dmitry.gumen...@gmail.com wrote:

> +@table @option
> +@item d
> +Dimensions @code{WxH}.
> +@code{W} - number of data values in json, values will be scaled according to 
> @code{H}.
> +The default value is @var{640x480}
> +
> +@item s
> +Samples count per value per channel

In most other filters/filtersources, s+size is used for dimensions,
d+duration for time, and n for an amount/number of frames/samples or
so. Might be a good idea to align with this. And use aliases (i.e.
implement both "s" and "size").

> +static const AVOption dumpwave_options[] = {
> +{ "d", "set width and height", OFFSET(w), AV_OPT_TYPE_IMAGE_SIZE, {.str 
> = "640x480"}, 0, 0, FLAGS },
> +{ "s", "set number of samples per value per channel",  OFFSET(s), 
> AV_OPT_TYPE_INT64,  {.i64 = 0}, 0, INT64_MAX, FLAGS },
> +{ "json", "set json dump file", OFFSET(json), AV_OPT_TYPE_STRING, { .str 
> = NULL }, 0, 0, FLAGS },
> +{ NULL }
[...]
> +static av_cold int init(AVFilterContext *ctx)
> +{
> +DumpWaveContext *dumpwave = ctx->priv;
> +if(!dumpwave->s) {
> +dumpwave->is_disabled = 1;
> +av_log(ctx, AV_LOG_ERROR, "Invalid samples per value config\n");
> +}
> +return 0;

I don't like the idea of having to provide the "s" parameter. Is there
really no useful default?

And now, if s=0, what does the filter do? Just sit around? Couldn't it
fail instead?

Apart from that:
"if (" is the bracket style ffmpeg uses.

> +if (dumpwave->json && !(dump_fp = av_fopen_utf8(dumpwave->json, 
> "w")))
> +av_log(ctx, AV_LOG_WARNING, "Flushing dump failed\n");

I would appreciate evaluation of errno and printing the appropriate
string (using av_strerror(), I believe).

> +switch (inlink->format) {
> +case AV_SAMPLE_FMT_DBLP:

As Kyle hinted: Can this not force a conversion (implicit insertion of
aformat filter) to e.g. double by only supporting this format, and
reducing this switch to one or two cases? (The filter's output isn't
really meant for transparent reuse anyway? af_volumedetect e.g. also
supports only one, meaning its output can be a different format than
its input.) It's also really hard to go through and later to maintain.
It the big switch/case remains, one or two code macros would be
appropriate.

> +for (i = 0; i < buf->nb_samples; i++) {
> +for (c = 0; c < channels; c++, src++)
> +calc_db_rms(dumpwave, *src);
> +}}

I believe the curly brackets need to be placed differently.

> +.description   = NULL_IF_CONFIG_SMALL("Dumps RMS amplitude to JSON 
> file"),

Imperative, i.e. "Dump RMS ...".

> +FATE_AFILTER-$(call FILTERDEMDEC, DUMPWAVE, WAV, PCM_S16LE) += 
> fate-filter-dumpwave
> +fate-filter-dumpwave: SRC = $(TARGET_PATH)/tests/data/asynth-44100-2.wav
> +fate-filter-dumpwave: CMD = ffmpeg -i $(SRC) -af 
> dumpwave=d=1800x140:s=147:json=$(TARGET_PATH)/tests/data/fate/filter-dumpwave.out
>  -f null - && cat $(TARGET_PATH)/tests/data/fate/filter-dumpwave.out
> +

If you need to implement all formats, you might want to test them all.
Just a thought.

Moritz
___
ffmpeg-devel mailing list
ffmpeg-devel@ffmpeg.org
http://ffmpeg.org/mailman/listinfo/ffmpeg-devel

Re: [FFmpeg-devel] [PATCH] avfilter: add dumpwave filter.

2018-01-10 Thread Tobias Rapp 

On 10.01.2018 09:34, Дмитрий Гуменюк wrote:

There is no existing dumping format for RMS. How to make it generic?


Some of my applications use the following pattern to dump audio envelope 
data during transcoding:


ffmpeg -i input-file -f [...] output-file \
  -f wav -write_peak only peak-file

The data format of "peak-file" follows EBU Tech 3285 Supplement 3 (BWF 
Peak Envelope Chunk).


Regards,
Tobias

___
ffmpeg-devel mailing list
ffmpeg-devel@ffmpeg.org
http://ffmpeg.org/mailman/listinfo/ffmpeg-devel

Re: [FFmpeg-devel] [PATCH] avfilter: add dumpwave filter.

2018-01-10 Thread Tobias Rapp 

On 10.01.2018 09:11, Дмитрий Гуменюк wrote:

Hi,
In my opinion JSON is more flexible, so envelope can be rendered in web app or 
mobile app
This filter is useful when you’d like to generate visual representation of 
audio while doing transcode


I see, but my point is: If you do the image rendering external anyway, 
why not use an existing envelope dumping format? Or alternatively 
connect the rendering application via the av* library interface in C?


Filters into the codebase should be as generic as possible to allow 
re-usage by other filters.





On 10 Jan 2018, at 09:00, Tobias Rapp  wrote:

On 08.01.2018 01:36, dmitry.gumen...@gmail.com wrote:

From: Dmytro Humeniuk 
Signed-off-by: Dmytro Humeniuk 
---
  Changelog  |   1 +
  libavfilter/Makefile   |   1 +
  libavfilter/af_dumpwave.c  | 273 +
  libavfilter/allfilters.c   |   1 +
  libavfilter/version.h  |   4 +-
  tests/fate/filter-audio.mak|   5 +
  tests/ref/fate/filter-dumpwave |   1 +
  7 files changed, 284 insertions(+), 2 deletions(-)
  create mode 100644 libavfilter/af_dumpwave.c
  create mode 100644 tests/ref/fate/filter-dumpwave
[...]


As far as I can see the filter reads audio and writes an envelope curve using 
JSON data format. The JSON data is then rendered into an image outside of 
FFmpeg.

In my opinion it would be better to allow connecting the filter with other filters by 
directly rendering the image within FFmpeg. So the filter should have a generic video 
output instead of the JSON output. This would be similar to the existing 
"showvolume" filter.

If you just want to get the raw envelope data from FFmpeg I suggest to take a look at the 
"write_peak" option of the WAV muxer.

Regards,
Tobias





BTW: Top-posting is unpopular on this list.

Regards,
Tobias

___
ffmpeg-devel mailing list
ffmpeg-devel@ffmpeg.org
http://ffmpeg.org/mailman/listinfo/ffmpeg-devel

Re: [FFmpeg-devel] [PATCH] avfilter: add dumpwave filter.

2018-01-10 Thread Tobias Rapp 

On 08.01.2018 01:36, dmitry.gumen...@gmail.com wrote:

From: Dmytro Humeniuk 

Signed-off-by: Dmytro Humeniuk 
---
  Changelog  |   1 +
  libavfilter/Makefile   |   1 +
  libavfilter/af_dumpwave.c  | 273 +
  libavfilter/allfilters.c   |   1 +
  libavfilter/version.h  |   4 +-
  tests/fate/filter-audio.mak|   5 +
  tests/ref/fate/filter-dumpwave |   1 +
  7 files changed, 284 insertions(+), 2 deletions(-)
  create mode 100644 libavfilter/af_dumpwave.c
  create mode 100644 tests/ref/fate/filter-dumpwave

[...]


As far as I can see the filter reads audio and writes an envelope curve 
using JSON data format. The JSON data is then rendered into an image 
outside of FFmpeg.


In my opinion it would be better to allow connecting the filter with 
other filters by directly rendering the image within FFmpeg. So the 
filter should have a generic video output instead of the JSON output. 
This would be similar to the existing "showvolume" filter.


If you just want to get the raw envelope data from FFmpeg I suggest to 
take a look at the "write_peak" option of the WAV muxer.


Regards,
Tobias

___
ffmpeg-devel mailing list
ffmpeg-devel@ffmpeg.org
http://ffmpeg.org/mailman/listinfo/ffmpeg-devel

Re: [FFmpeg-devel] [PATCH] avfilter: add dumpwave filter.

2018-01-09 Thread Dmitry Gumenyuk 
Hi,
>Can you provide some context for this? Please add docs as well.
filter converts samples to decibels and calculates RMS
(Root-Mean-Square) audio power scaled to desired values,
it supposed to generate something like SoundCloud does
http://plnkr.co/edit/WeppGcEPQiOyKDsA0C4Y
I added docs.
>Can't really test this because it segfaults for me:  ./ffmpeg -f lavfi
>-i anoisesrc -af dumpwave -f null -
it expected some config - fixed

>Just set the sample format to something like AV_SAMPLE_FMT_DBL in
>query_format. There's no need for this big switch statement.
it handles samples in different way, you may look at the astats filter

2018-01-08 23:58 GMT+01:00 Kyle Swanson :
> Hi,
>
> Can you provide some context for this? Please add docs as well.
> Can't really test this because it segfaults for me:  ./ffmpeg -f lavfi
> -i anoisesrc -af dumpwave -f null -
>
> On Sun, Jan 7, 2018 at 4:36 PM,  wrote:
>>
>> From: Dmytro Humeniuk 
>>
>> Signed-off-by: Dmytro Humeniuk 
>> ---
>>  Changelog  |   1 +
>>  libavfilter/Makefile   |   1 +
>>  libavfilter/af_dumpwave.c  | 273 
>> +
>>  libavfilter/allfilters.c   |   1 +
>>  libavfilter/version.h  |   4 +-
>>  tests/fate/filter-audio.mak|   5 +
>>  tests/ref/fate/filter-dumpwave |   1 +
>>  7 files changed, 284 insertions(+), 2 deletions(-)
>>  create mode 100644 libavfilter/af_dumpwave.c
>>  create mode 100644 tests/ref/fate/filter-dumpwave
>>
>> diff --git a/Changelog b/Changelog
>> index 61075b3392..40fd624449 100644
>> --- a/Changelog
>> +++ b/Changelog
>> @@ -38,6 +38,7 @@ version :
>>  - Removed the ffserver program
>>  - Removed the ffmenc and ffmdec muxer and demuxer
>>  - VideoToolbox HEVC encoder and hwaccel
>> +- dumpwave audio filter
>>
>>
>>  version 3.4:
>> diff --git a/libavfilter/Makefile b/libavfilter/Makefile
>> index 256dfabd66..020d90ee01 100644
>> --- a/libavfilter/Makefile
>> +++ b/libavfilter/Makefile
>> @@ -122,6 +122,7 @@ OBJS-$(CONFIG_TREMOLO_FILTER)+= 
>> af_tremolo.o
>>  OBJS-$(CONFIG_VIBRATO_FILTER)+= af_vibrato.o 
>> generate_wave_table.o
>>  OBJS-$(CONFIG_VOLUME_FILTER) += af_volume.o
>>  OBJS-$(CONFIG_VOLUMEDETECT_FILTER)   += af_volumedetect.o
>> +OBJS-$(CONFIG_DUMPWAVE_FILTER)   += af_dumpwave.o
>>
>>  OBJS-$(CONFIG_AEVALSRC_FILTER)   += aeval.o
>>  OBJS-$(CONFIG_ANOISESRC_FILTER)  += asrc_anoisesrc.o
>> diff --git a/libavfilter/af_dumpwave.c b/libavfilter/af_dumpwave.c
>> new file mode 100644
>> index 00..493b5b7ff2
>> --- /dev/null
>> +++ b/libavfilter/af_dumpwave.c
>> @@ -0,0 +1,273 @@
>> +/*
>> + * Copyright (c) 2017 Dmytro Humeniuk
>> + *
>> + * This file is part of FFmpeg.
>> + *
>> + * FFmpeg is free software; you can redistribute it and/or
>> + * modify it under the terms of the GNU Lesser General Public
>> + * License as published by the Free Software Foundation; either
>> + * version 2.1 of the License, or (at your option) any later version.
>> + *
>> + * FFmpeg is distributed in the hope that it will be useful,
>> + * but WITHOUT ANY WARRANTY; without even the implied warranty of
>> + * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU
>> + * Lesser General Public License for more details.
>> + *
>> + * You should have received a copy of the GNU Lesser General Public
>> + * License along with FFmpeg; if not, write to the Free Software
>> + * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 
>> USA
>> + */
>> +
>> +/**
>> + * @file
>> + * waveform audio filter – dumps RMS amplitude to JSON file like SoundCloud 
>> does
>> + */
>> +
>> +#include "libavutil/avassert.h"
>> +#include "libavutil/avstring.h"
>> +#include "libavutil/channel_layout.h"
>> +#include "libavutil/opt.h"
>> +#include "libavutil/parseutils.h"
>> +#include "avfilter.h"
>> +#include "formats.h"
>> +#include "audio.h"
>> +#include "internal.h"
>> +
>> +typedef struct DumpWaveContext {
>> +const AVClass *class;
>> +int w, h;
>> +AVRational rate;
>> +int col;
>> +char *json;
>> +char *str;
>> +double *values;
>> +int64_t c, n, max_samples;
>> +double sum; /* sum of the squared samples per segment */
>> +} DumpWaveContext;
>> +
>> +#define OFFSET(x) offsetof(DumpWaveContext, x)
>> +#define FLAGS AV_OPT_FLAG_AUDIO_PARAM|AV_OPT_FLAG_FILTERING_PARAM
>> +
>> +static const AVOption dumpwave_options[] = {
>> +{ "s","set dump size", OFFSET(w), AV_OPT_TYPE_IMAGE_SIZE, {.str = 
>> "600x240"}, 0, 0, FLAGS },
>> +{ "c", "set number of samples per item",  OFFSET(c), AV_OPT_TYPE_INT64, 
>>  {.i64 = 0}, 0, INT64_MAX, FLAGS },
>
> Use more descriptive parameter names. Also, per item? What is an item?
>
>> +{ "json", "set dump file", OFFSET(json), AV_OPT_TYPE_STRING, { .str = 
>> NULL }, 0, 0, FLAGS },
>> +{ NULL

Re: [FFmpeg-devel] [PATCH] avfilter: add dumpwave filter.

2018-01-08 Thread Kyle Swanson 
Hi,

Can you provide some context for this? Please add docs as well.
Can't really test this because it segfaults for me:  ./ffmpeg -f lavfi
-i anoisesrc -af dumpwave -f null -

On Sun, Jan 7, 2018 at 4:36 PM,  wrote:
>
> From: Dmytro Humeniuk 
>
> Signed-off-by: Dmytro Humeniuk 
> ---
>  Changelog  |   1 +
>  libavfilter/Makefile   |   1 +
>  libavfilter/af_dumpwave.c  | 273 
> +
>  libavfilter/allfilters.c   |   1 +
>  libavfilter/version.h  |   4 +-
>  tests/fate/filter-audio.mak|   5 +
>  tests/ref/fate/filter-dumpwave |   1 +
>  7 files changed, 284 insertions(+), 2 deletions(-)
>  create mode 100644 libavfilter/af_dumpwave.c
>  create mode 100644 tests/ref/fate/filter-dumpwave
>
> diff --git a/Changelog b/Changelog
> index 61075b3392..40fd624449 100644
> --- a/Changelog
> +++ b/Changelog
> @@ -38,6 +38,7 @@ version :
>  - Removed the ffserver program
>  - Removed the ffmenc and ffmdec muxer and demuxer
>  - VideoToolbox HEVC encoder and hwaccel
> +- dumpwave audio filter
>
>
>  version 3.4:
> diff --git a/libavfilter/Makefile b/libavfilter/Makefile
> index 256dfabd66..020d90ee01 100644
> --- a/libavfilter/Makefile
> +++ b/libavfilter/Makefile
> @@ -122,6 +122,7 @@ OBJS-$(CONFIG_TREMOLO_FILTER)+= 
> af_tremolo.o
>  OBJS-$(CONFIG_VIBRATO_FILTER)+= af_vibrato.o 
> generate_wave_table.o
>  OBJS-$(CONFIG_VOLUME_FILTER) += af_volume.o
>  OBJS-$(CONFIG_VOLUMEDETECT_FILTER)   += af_volumedetect.o
> +OBJS-$(CONFIG_DUMPWAVE_FILTER)   += af_dumpwave.o
>
>  OBJS-$(CONFIG_AEVALSRC_FILTER)   += aeval.o
>  OBJS-$(CONFIG_ANOISESRC_FILTER)  += asrc_anoisesrc.o
> diff --git a/libavfilter/af_dumpwave.c b/libavfilter/af_dumpwave.c
> new file mode 100644
> index 00..493b5b7ff2
> --- /dev/null
> +++ b/libavfilter/af_dumpwave.c
> @@ -0,0 +1,273 @@
> +/*
> + * Copyright (c) 2017 Dmytro Humeniuk
> + *
> + * This file is part of FFmpeg.
> + *
> + * FFmpeg is free software; you can redistribute it and/or
> + * modify it under the terms of the GNU Lesser General Public
> + * License as published by the Free Software Foundation; either
> + * version 2.1 of the License, or (at your option) any later version.
> + *
> + * FFmpeg is distributed in the hope that it will be useful,
> + * but WITHOUT ANY WARRANTY; without even the implied warranty of
> + * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU
> + * Lesser General Public License for more details.
> + *
> + * You should have received a copy of the GNU Lesser General Public
> + * License along with FFmpeg; if not, write to the Free Software
> + * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 
> USA
> + */
> +
> +/**
> + * @file
> + * waveform audio filter – dumps RMS amplitude to JSON file like SoundCloud 
> does
> + */
> +
> +#include "libavutil/avassert.h"
> +#include "libavutil/avstring.h"
> +#include "libavutil/channel_layout.h"
> +#include "libavutil/opt.h"
> +#include "libavutil/parseutils.h"
> +#include "avfilter.h"
> +#include "formats.h"
> +#include "audio.h"
> +#include "internal.h"
> +
> +typedef struct DumpWaveContext {
> +const AVClass *class;
> +int w, h;
> +AVRational rate;
> +int col;
> +char *json;
> +char *str;
> +double *values;
> +int64_t c, n, max_samples;
> +double sum; /* sum of the squared samples per segment */
> +} DumpWaveContext;
> +
> +#define OFFSET(x) offsetof(DumpWaveContext, x)
> +#define FLAGS AV_OPT_FLAG_AUDIO_PARAM|AV_OPT_FLAG_FILTERING_PARAM
> +
> +static const AVOption dumpwave_options[] = {
> +{ "s","set dump size", OFFSET(w), AV_OPT_TYPE_IMAGE_SIZE, {.str = 
> "600x240"}, 0, 0, FLAGS },
> +{ "c", "set number of samples per item",  OFFSET(c), AV_OPT_TYPE_INT64,  
> {.i64 = 0}, 0, INT64_MAX, FLAGS },

Use more descriptive parameter names. Also, per item? What is an item?

> +{ "json", "set dump file", OFFSET(json), AV_OPT_TYPE_STRING, { .str = 
> NULL }, 0, 0, FLAGS },
> +{ NULL }
> +};
> +
> +AVFILTER_DEFINE_CLASS(dumpwave);
> +
> +static int config_output(AVFilterLink *outlink)
> +{
> +DumpWaveContext *dumpwave = outlink->src->priv;
> +const int ch_width = dumpwave->w;
> +dumpwave->values = av_malloc(ch_width * sizeof(double));
> +dumpwave->str = av_malloc(ch_width*sizeof(int));
> +dumpwave->max_samples = dumpwave->c * outlink->channels;
> +
> +return 0;
> +}
> +
> +static av_cold void uninit(AVFilterContext *ctx)
> +{
> +DumpWaveContext *dumpwave = ctx->priv;
> +const int ch_height = dumpwave->h;
> +const int ch_width = dumpwave->w;
> +char *result = dumpwave->str;
> +FILE *dump_fp = NULL;
> +
> +if (dumpwave->json && !(dump_fp = av_fopen_utf8(dumpwave->json, "w")))
> +av_log(ctx, AV_LOG_WARNING, "dump failed.\n");
> +