Post-processing techniques used before saving the call recordings?

hamza1582 · January 18, 2023, 8:50am

Hello there!
I have observed that my FreePBX server is receiving the audio data in G.711-PCMU(bit-depth of 8) codec through SIP Trunk connection. But it is saving the call recordings with bith-depth of 16 and it is quite clear and smooth.
So, I want to know what are the post-processing techniques it is using to upscale the audio and where the script for this post-processing is actually implemented?
I will be thankful for any help and guidance.

david55 · January 18, 2023, 2:08pm

github.com

asterisk/asterisk/blob/master/codecs/codec_ulaw.c#L45


      
          #include "asterisk/ulaw.h"
          #include "asterisk/utils.h"
          
          
#define BUFFER_SAMPLES   8096	/* size for the translation buffers */
          
          
/* Sample frame data */
          #include "asterisk/slin.h"
          #include "ex_ulaw.h"
          
          
/*! \brief convert and store samples in outbuf */
          static int ulawtolin_framein(struct ast_trans_pvt *pvt, struct ast_frame *f)
          {
          	int i = f->samples;
          	unsigned char *src = f->data.ptr;
          	int16_t *dst = pvt->outbuf.i16 + pvt->samples;
          
          
	pvt->samples += i;
          	pvt->datalen += i * 2;	/* 2 bytes/sample */
          
          
	/* convert and copy in outbuf */
          	while (i--)

which is a simple table lookup, which is the obvious way of doing it.

More generally, Asterisk will use the same code it uses to transcode between channels.

Note that G.711 is not linear. It covers the full 16 dynamic range.

system · January 25, 2023, 2:08pm

This topic was automatically closed 7 days after the last reply. New replies are no longer allowed.