[asterisk-dev] adjusting the playback speed of voicemail messages.

[John Todd]

At 4:33 PM -0500 4/2/08, Steven S. Critchfield wrote:
>----- "Travis Hein" <travis.hein at travnet.org> wrote:
>  > I'm hoping I'm missing something and there is some easy way, but
>  > after
>>  looking at the code, it appears that the sample rate for the ast_frame
>>  is
>>  assumed to be 8 kHz. The formats/format_wav.c seems to also assume 8
>>  kHz
>>  always as well, and the ast_format does not seem to have an
>>   int (*set_sample_rate)(struct ast_filestream *, int new_rate);
>>  either, which I was thinking would need something like that so that
>>  the
>>  ast_control_streamfile() (that is called by the voicemail playback)
>>  could
>>  adjust the sample rate by a given amount, just like it currently does
>>  for
>>  the skipms value to the seek() function.
>
>The problem with speed up and slow down is that just adjusting the 
>sample rate doesn't make inteligent sounding audio. I'm sure you are 
>familiar with the sound of audio from a tape recorder when you also 
>hit the fast forward button. Sounds a lot like chipmunks.
>
>Also the sample rates are set specifically because that is what it 
>takes to get it onto PSTN networks. I believe there is another 
>branch where they have worked on that assumption, but it isn't the 
>trunk from what I know.
>
>What you need to be able to do is essentially set up either a filter 
>or a temporary file. A filter would sit in the middle of the audio 
>stream and apply some tricks to properly stretch or compress the 
>audio in such a way as to not affect pitch. The temp file route 
>would be to use sox to modify the audio file and maintain the sample 
>rate and such so you could just work from the modified audio file.
>
>Downside is, I don't think putting a audio filter in the loop is 
>currently feasible in the trunk, again, other branches I believe are 
>working on this. I don't know how easy the sox route would be for 
>you. Especially if you expect to maintain a pointer into the audio 
>file and only effectively adjust from that spot forward. or restart 
>from the sport you stopped to modify the audio file.
>--
>Steven Critchfield critch at basesys.com


I believe that 1.6 branches include an interface to Jack... let's see... yes.

http://svn.digium.com/view/asterisk/branches/1.6.0/CHANGES?view=markup

   502   * Added a new module, app_jack, which provides interfaces to 
JACK, the Jack
   503      Audio Connection Kit (http://www.jackaudio.org/).  Two 
interfaces are
   504      provided; there is a JACK() application, and a JACK_HOOK() 
function.  Both
   505      interfaces create an input and output JACK port.  The 
application makes
   506      these ports the endpoint of the call.  The audio coming 
from the channel
   507      goes out the output port and whatever comes back in on the 
input port is
   508      what gets sent to the channel.  The JACK_HOOK() function 
turns on a JACK
   509      audiohook on the channel.  This lets you run the audio coming from a
   510      channel through JACK, and whatever comes back in is what 
gets forwarded
   511      on as the channel's audio.  This is very useful for building custom
   512      vocoders or doing recording or analysis of the channel's 
audio in another
   513      application.



This most likely can do what you want, but I don't have a clear 
understanding of the interface method at this point, and I don't know 
enough about how JACK works to say what a filter would look like to 
slow down or speed up audio without distorting the pitch.  Let us 
know if this is the right solution, and the methods and details of 
any successful tests!

JT