In this case no news = bad news ...
Based on Mario´s post and the possibilities I could think about in the
last days I have modified the code trying to isolate the issues like
calling blocking methods and as a consequence loosing part of the
audio stream, having buffer overflows, etc, as this could be the
reason for the bad audio playback.
Here is the code:
package com.example.test;
import android.app.Activity;
import android.media.AudioFormat;
import android.media.AudioManager;
import android.media.AudioRecord;
import android.media.AudioTrack;
import android.media.MediaRecorder;
import android.os.Bundle;
public class test extends Activity {
/** Called when the activity is first created. */
@Override
public void onCreate(Bundle savedInstanceState) {
super.onCreate(savedInstanceState);
setContentView(R.layout.main);
Record record = new Record();
record.run();
}
public class Record extends Thread
{
public void run() {
android.os.Process.setThreadPriority
(android.os.Process.THREAD_PRIORITY_URGENT_AUDIO);
AudioRecord arec = new
AudioRecord(MediaRecorder.AudioSource.MIC,
8000,
AudioFormat.CHANNEL_CONFIGURATION_MONO,
AudioFormat.ENCODING_PCM_16BIT,
32000);
AudioTrack atrack = new
AudioTrack(AudioManager.STREAM_MUSIC,
8000,
AudioFormat.CHANNEL_CONFIGURATION_MONO,
AudioFormat.ENCODING_PCM_16BIT,
56000,
AudioTrack.MODE_STREAM);
short[] buffer = new short[64000];
int t = 0;
arec.startRecording();
while(t < 64000)
t += arec.read(buffer, t, 8000);
arec.stop();
arec.release();
t = 0;
atrack.play();
while(t < 64000)
t += atrack.write(buffer, t, 8000);
atrack.stop();
atrack.release();
}
}
}
As you can see, first I record the audio and just after stop recording
I start playing it. As expected there was no AudioRecord buffer
overflow messages and I´ve indexed the buffers by the return values
received by calling read() and write().
The sonic result was the same: chopped audio with a sensation of a
slow playback, and distorted. The impression I have also is that it´s
not recording for the whole 6 seconds, or at least the playback doesn
´t last it all.
Could anyone try this simple code on a real device and check if it
works ok?
Thanks,
Gabriel Simões
On 17 mar, 22:58, Gabriel Simões <[email protected]> wrote:
> Thanks for your answer Mario!
>
> Yes, I wrote it wrong ... should have been AudioTrack.write() instead
> of AudioTrack.play().
> The same idea came to my mind a couple of hours ago. Since AudioTrack
> when set
> as stream will only start playing at the time it´s internal buffer
> gets full, write()
> could block if the buffer is full to prevent an overflow.
>
> To quick test it, I have instantiated AudioTrack with an internal
> buffer a lot
> bigger than the one I used to read and write the audio streams. This
> way my app should
> take longer to start playing and get a pretty big delay, but the
> audio
> should not chop nor present any artfacts at least for the first
> seconds.
>
> The results I got presented the same problems, but based on that I
> will try it again.
>
> One other thing that came to my mind is, once a buffer overflow
> happens, does AudioRecord handle the problem and reorganize the buffer
> or
> should I flush it and start from 0 (the buffer gets messed and this
> ends giving a bad audio signal with chops, delay, etc...)?
>
> Gabriel
>
> On 17 mar, 21:07, Mario Zechner <[email protected]> wrote:
>
>
>
> > Hi there,
>
> > not sure whether i can help but you stated that you couldn't find
> > anywhere that "AudioTrack.play()" (which should probably read
> > AudioTrack.write()) does not block. Well, it does. And if you do the
> > following in your thread:
>
> > AudioRecord.read(sampels)
> > AudioTrack.write(samples)
>
> > i can imagine that the write will block long enough for the
> > AudioRecord to overflow as you don't read in the next buffer of
> > samples early enough. You could try decreasing the number of samples
> > you read and write so the write will not block as long. No idea
> > whether that will solve the problem.
>
> > hth,
> > Mario
>
> > On 18 Mrz., 00:33, HeHe <[email protected]> wrote:
>
> > > no, i havent, because my app runs ok on g1 with 1.6 firmware. it just
> > > runs weird on 1.6 emulator. so i am 90% happy already.
>
> > > btw, my app does not record and play audio at the same time, but it
> > > encodes audio with speex, which should take overhead no less than
> > > playback, i guess.
>
> > > anyway, i wish you lucks and hope goodle folks will help you solve the
> > > issue.
>
> > > On Mar 17, 8:55 am, Gabriel Simões <[email protected]> wrote:
>
> > > > Installing ubuntu 9.1 on my machine.
> > > > I will try using linux to develop as I think it will be less confusing
> > > > to develop using NDK.
>
> > > > Hehe: I was thinking about trying to get an answer from someone from
> > > > Google tomorrow at irc. Have you ever tried it?
>
> > > > Won´t anyone who is actually using AudioRecord with success show up
> > > > and tell us what we are doing wrong? :(
>
> > > > On 17 mar, 12:53, HeHe <[email protected]> wrote:
>
> > > > > i heard that too. but some google folk seemed discouraging use of the
> > > > > stuff.
>
> > > > > On Mar 17, 6:27 am, Gabriel Simões <[email protected]> wrote:
>
> > > > > > Nope,
>
> > > > > > I´ve paid attention to see if GC was being called, but nothing shows
> > > > > > up on LogCat.
> > > > > > Well, I´ve found some posts on the internet about using
> > > > > > AudioRecord´s
> > > > > > native implementation. Isn´t it supported anymore?
>
> > > > > > I still can´t believe we are spending so much time on something that
> > > > > > should be "plug and play". It´s almost becoming "plug and pray"!
>
> > > > > > On 17 mar, 02:41, HeHe <[email protected]> wrote:
>
> > > > > > > if you find any native audio interfaces in NDK, please share it
> > > > > > > with
> > > > > > > me first :)
>
> > > > > > > for your issue, did you check GC? that is, around the time when
> > > > > > > you
> > > > > > > see buffer overflow msgs, does GC happen too?
>
> > > > > > > On Mar 16, 7:33 pm, Gabriel Simões <[email protected]> wrote:
>
> > > > > > > > *** Idea ...
> > > > > > > > Using the NDK native audio interfaces should solve this problem?
>
> > > > > > > > On 16 mar, 22:26, Gabriel Simões <[email protected]> wrote:
>
> > > > > > > > > News, but no good news...
>
> > > > > > > > > I´ve tried the source code above on another machine running
> > > > > > > > > the sdk
> > > > > > > > > 2.1, an AVD with default parameters plus audio record and
> > > > > > > > > audio
> > > > > > > > > playback capabilities, another operational system (linux),
> > > > > > > > > etc.
> > > > > > > > > The results? the same we had before .... audio delay (long
> > > > > > > > > delay),
> > > > > > > > > chopping, distortion and a metalic voice.
>
> > > > > > > > > I can not believe nobody here in this forum hasn´t been able
> > > > > > > > > to
> > > > > > > > > develop any app that can record from the microfone using
> > > > > > > > > AudioRecord.
> > > > > > > > > Also, anyone here hasn´t even made a loop back from microfone
> > > > > > > > > to line
> > > > > > > > > out to check the audio being recorded?
>
> > > > > > > > > Please, c´mon guys ... this is not a commercial secret, this
> > > > > > > > > won´t
> > > > > > > > > increase competition, and I really can´t see any other
> > > > > > > > > reasons not to
> > > > > > > > > see some posts from developers who work with audio or other
> > > > > > > > > signal
> > > > > > > > > processing using Android.
>
> > > > > > > > > Please, show us a way to go because I´ve tried almost all the
> > > > > > > > > possibilities here (the only one left is to try using a real
> > > > > > > > > device)
> > > > > > > > > and revising the code I can´t find what else it could be.
>
> > > > > > > > > *Obs: once again, a program with a thread only to read from
> > > > > > > > > AudioRecord and play with AudioTrack is giving warnings about
> > > > > > > > > buffer
> > > > > > > > > overflows (Audio Record). There´s no processing going on and
> > > > > > > > > since I
> > > > > > > > > haven´t found anything related to "AudioTrack.play() blocks
> > > > > > > > > until the
> > > > > > > > > audio stream is played", based on it´s actuall behavior in my
> > > > > > > > > app I can
> > > > > > > > > ´t see how this class could be useful
> > > > > > > > > for online audio processing applications.
>
> > > > > > > > > Thank you,
> > > > > > > > > Gabriel
>
> > > > > > > > > On 15 mar, 20:53, Gabriel Simões <[email protected]> wrote:
>
> > > > > > > > > > Well, just to try I´ve downloaded the SDK 1.6 (v. 4),
> > > > > > > > > > created a new
> > > > > > > > > > AVD based on that and then tried the source code from the
> > > > > > > > > > first post
> > > > > > > > > > in two different situations:
>
> > > > > > > > > > - App compiled using SDK 1.6 on an AVD running 1.6
> > > > > > > > > > - App compiled using SDK 1.6 on an AVD running 2.1
>
> > > > > > > > > > Had exactly the same results as "compiling" and running on
> > > > > > > > > > 2.1: audio
> > > > > > > > > > chopping, distortion and artfacts even if I solo the
> > > > > > > > > > microfone input
> > > > > > > > > > level (using a 1 second buffer, sometimes the audio decays
> > > > > > > > > > in a lot
> > > > > > > > > > longer) and AudioRecord buffer overflow messages even if
> > > > > > > > > > all I do in
> > > > > > > > > > the thread is to read from AudioRecord and to play using
> > > > > > > > > > AudioTrack.
>
> > > > > > > > > > Tomorrow I will try to test my code using another computer
> > > > > > > > > > and even
> > > > > > > > > > on
> > > > > > > > > > a real device. In the mean time, any other ideas?
>
> > > > > > > > > > Gabriel
>
> > > > > > > > > > On 15 mar, 01:31, HeHe <[email protected]> wrote:
>
> > > > > > > > > > > no chopping, likely because my computer (Core2 E8400) is
> > > > > > > > > > > lightening
> > > > > > > > > > > fast.
>
> > > > > > > > > > > my AudioRecord.read() returns a value equal to my buffer
> > > > > > > > > > > size, too.
>
> > > > > > > > > > > however, it returns recorded samples in a speed "faster
> > > > > > > > > > > than 8khz",
> > > > > > > > > > > which makes later playback sound like my voice is slowed.
>
> > > > > > > > > > > again, the same code works correctly on 1.5 emulator.
>
> > > > > > > > > > > so, seems your problem is not the same as mine :(
>
> > > > > > > > > > > On Mar 14, 6:53 pm, Gabriel Simões <[email protected]>
> > > > > > > > > > > wrote:
>
> > > > > > > > > > > > Audio chopping also?
> > > > > > > > > > > > Could you please test the code above?
>
> > > > > > > > > > > > I´ve checked and AudioRecord.read() returns a value
> > > > > > > > > > > > equal to my buffer
> > > > > > > > > > > > size. Based on that I understand that
> > > > > > > > > > > > AudioRecord.read() blocks the
> > > > > > > > > > > > execution, waits for the buffer to get a number of
> > > > > > > > > > > > samples at least
> > > > > > > > > > > > equal to the buffer size and then it returns the audio
> > > > > > > > > > > > stream (older
> > > > > > > > > > > > "buffer size" number of samples in the AudioRecord
> > > > > > > > > > > > internal buffer).
>
> > > > > > > > > > > > On 14 mar, 22:34, HeHe <[email protected]> wrote:
>
> > > > > > > > > > > > > seems i am experiencing similar issue as you
> > > > > > > > > > > > > r.http://groups.google.com/group/android-developers/browse_thread/threa...
>
> > > > > > > > > > > > > without any single line of code modification to my
> > > > > > > > > > > > > app, the count of
> > > > > > > > > > > > > frames recorded by app running on sdk 1.6 emulator is
> > > > > > > > > > > > > significantly
> > > > > > > > > > > > > more than that recorded by the same app running on
> > > > > > > > > > > > > sdk 1.5 emulator.
>
> > > > > > > > > > > > > however, the same app runs equally well on both
> > > > > > > > > > > > > 1.5/1.6 emulators.
>
> > > > > > > > > > > > > it is weird.....
>
> > > > > > > > > > > > > On Mar 14, 3:00 pm, Gabriel Simões
> > > > > > > > > > > > > <[email protected]> wrote:
>
> > > > > > > > > > > > > > Hello,
>
> > > > > > > > > > > > > > Since my day 1 I´ve been facing problems with
> > > > > > > > > > > > > > Android AudioRecord API.
> > > > > > > > > > > > > > After a lot of research and a lot more trial and
> > > > > > > > > > > > > > error I´ve found my
> > > > > > > > > > > > > > way into the code and was able to write an
> > > > > > > > > > > > > > asynctask for audio streams
> > > > > > > > > > > > > > recording which encapsules most of the audiorecord
> > > > > > > > > > > > > > and asynctask
> > > > > > > > > > > > > > features as a lib, so the developers doesn´t need
> > > > > > > > > > > > > > to worry about
> > > > > > > > > > > > > > finding the best audio parameters, work on stop and
> > > > > > > > > > > > > > release, etc.
> > > > > > > > > > > > > > So far, everything was right.
> > > > > > > > > > > > > > Yesterday I started trying to process the audio I
> > > > > > > > > > > > > > got from the
> > > > > > > > > > > > > > microfone input and my code was reacting bad. I
> > > > > > > > > > > > > > revisited it many
> > > > > > > > > > > > > > times and decided then to check the audio I was
> > > > > > > > > > > > > > getting from the mic.
> > > > > > > > > > > > > > Created in instance of AudioTrack and tried to play
> > > > > > > > > > > > > > it back. The sound
> > > > > > > > > > > > > > was horrible, with lots of chopping, lots of
> > > > > > > > > > > > > > distortion and giving me
> > > > > > > > > > > > > > even the sensation of low samplerate playback (a
> > > > > > > > > > > > > > single word- Ocultar texto das mensagens anteriores
> > > > > > > > > > > > > > -
>
> - Mostrar texto das mensagens anteriores -...
>
> mais »
--
You received this message because you are subscribed to the Google
Groups "Android Developers" group.
To post to this group, send email to [email protected]
To unsubscribe from this group, send email to
[email protected]
For more options, visit this group at
http://groups.google.com/group/android-developers?hl=en
To unsubscribe from this group, send email to
android-developers+unsubscribegooglegroups.com or reply to this email with the
words "REMOVE ME" as the subject.
