New submission from Alex Converse <alex.conve...@gmail.com>:
am05_96 does not produce accurate output.
The rest of the am05_* series works fine. The other 96 kHz streams: am0[0-4]_96
also work fine
ssnrcd output:
ISO/IEC 14496-4 MPEG-4 Conformance Verification
SSNRCD - Segmental SNR / Cepstral Distortion Analysis Tool version 2.0
=== ref2.wav and ff.wav:
AFrdWVhead - WAVE file: Header stucture size mismatch
channel: 0
samples compared : 287744
number of different samples : 1056
max. abs. difference sample : 1.43e-01 (threshold:
1.22e-04, 15 bits res.) -> FAIL
overall RMS value of difference : -66.51 dB (threshold: -95.08
dB, 15 bits res.) -> FAIL
average segmental SNR value of difference : 99.72 dB (320 samples per
segment) -> OK
cepstral distortion : 0.01 dB (320 samples per
segment) -> OK
channel: 1
samples compared : 287744
number of different samples : 1062
max. abs. difference sample : 1.56e-01 (threshold:
1.22e-04, 15 bits res.) -> FAIL
overall RMS value of difference : -65.76 dB (threshold: -95.08
dB, 15 bits res.) -> FAIL
average segmental SNR value of difference : 100.18 dB (320 samples per
segment) -> OK
cepstral distortion : 0.00 dB (320 samples per
segment) -> OK
channel: 2
samples compared : 287744
number of different samples : 1094
max. abs. difference sample : 1.56e-01 (threshold:
1.22e-04, 15 bits res.) -> FAIL
overall RMS value of difference : -65.76 dB (threshold: -95.08
dB, 15 bits res.) -> FAIL
average segmental SNR value of difference : 100.17 dB (320 samples per
segment) -> OK
cepstral distortion : 0.02 dB (320 samples per
segment) -> OK
channel: 3
samples compared : 287744
number of different samples : 22
max. abs. difference sample : 3.05e-05 (threshold:
1.22e-04, 15 bits res.) -> OK
overall RMS value of difference : -131.47 dB (threshold:
-95.08 dB, 15 bits res.) -> OK
average segmental SNR value of difference : 96.29 dB (320 samples per
segment) -> OK
cepstral distortion : 0.00 dB (320 samples per
segment) -> OK
channel: 4
samples compared : 287744
number of different samples : 1105
max. abs. difference sample : 1.56e-01 (threshold:
1.22e-04, 15 bits res.) -> FAIL
overall RMS value of difference : -65.76 dB (threshold: -95.08
dB, 15 bits res.) -> FAIL
average segmental SNR value of difference : 99.94 dB (320 samples per
segment) -> OK
cepstral distortion : 0.02 dB (320 samples per
segment) -> OK
channel: 5
samples compared : 287744
number of different samples : 1081
max. abs. difference sample : 1.56e-01 (threshold:
1.22e-04, 15 bits res.) -> FAIL
overall RMS value of difference : -65.76 dB (threshold: -95.08
dB, 15 bits res.) -> FAIL
average segmental SNR value of difference : 100.60 dB (320 samples per
segment) -> OK
cepstral distortion : 0.02 dB (320 samples per
segment) -> OK
=== Test on RMS and abs. max. difference sample : FAIL
=== Test on seg. SNR and cepstral distortion : OK
=== Reached RMS criterion for a maximum of : 10 bit
=== Reached LSB criterion for a maximum of : 4 bit
=== Reached RMS/LSB criteria for a maximum of : 4 bit
Debug output from the first frame:
Conformance Vector am05_96
debug option v enabled
ele_id 0
tag 0, common window 0
nch 1, channels 0 0, widx 0 0
win 0, wsh 0
max_sf 40
-DV
ele_id 1
win 0, wsh 0
max_sf 40
tag 0, common window 1
nch 2, channels 1 2, widx 1 1
ele_id 1
win 0, wsh 0
max_sf 40
tag 1, common window 1
nch 2, channels 33 34, widx 33 33
ele_id 3
tag 0, common window 0
nch 1, channels 48 48, widx 48 48
win 0, wsh 0
max_sf 40
ele_id 2
win 0, wsh 0
max_sf 40
ele_id 6
ele_id 6
ele_id 6
ele_id 7
sampling_rate_idx: 0
sampling_rate : 96000
profile: 0
Ch Info
idx tag cpe wdx cch dom ind
0 0 0 0 0 1 1
1 0 1 1 0 1 1
2 0 1 1 0 1 1
33 1 1 33 0 1 1
34 1 1 33 0 1 1
48 0 0 48
warning: no standards for 6 channels in one file... using multiple
Tool summary sort'd and uniq'd:
debug option V enabled
# DRC IS MS detected
# DRC IS MS TNS detected
# DRC IS Pred MS detected
# DRC IS Pred MS TNS detected
End of bitstream
The first few frames are:
# DRC IS MS detected
# DRC IS MS TNS detected
# DRC IS MS TNS detected
# DRC IS MS TNS detected
Difference waveform:
http://filer.case.edu/ajc30/pub/ffmpeg/am05_96_ch0_err.png
----------
messages: 4233
priority: normal
status: new
substatus: new
title: AAC: am05_96 does not produce accurate output
type: bug
____________________________________________________
FFmpeg issue tracker <ffmpeg_iss...@live.polito.it>
<https://roundup.ffmpeg.org/roundup/ffmpeg/issue838>
____________________________________________________
