[Freeswitch-users] [VBR]: Asynchronous PTIME supported on Speex16

[Stephen Dame]

Have an application that uses SPEEX at 20ms@16000,  everything works fine in
1.4, 1.5 and FreeSWITCH Version 1.7.0+git~20160219T153438Z~3bd26eaa6b~64bit
(git 3bd26ea 2016-02-19 15:34:38Z 64bit)

 

We bring a user into echo, then transfer them to a conference after they
confirm they can hear themselves.   We connect to audio fine at the 20ms and
confirm.   But the transfer is setting on VBR since updating freeswitch?

 

I built FreeSWITCH Version 1.7.0+git~20160706T181946Z~8c6b2657bf~64bit (git
8c6b265 2016-07-06 18:19:46Z 64bit)

 

2016-08-03 19:34:43.901960 [DEBUG] switch_rtp.c:6711 Correct audio ip/port
confirmed.

2016-08-03 19:34:43.901960 [WARNING] switch_core_media.c:2568 [VBR]:
Asynchronous PTIME supported, adjusting JB size. Remote PTIME changed from
[20] to [36]

2016-08-03 19:34:43.921964 [NOTICE] switch_core_media.c:2977 Deactivating
write resampler

2016-08-03 19:34:43.921964 [DEBUG] switch_core_media.c:2984 Changing Codec
from SPEEX at 20ms@16000hz to SPEEX at 36ms@16000hz

2016-08-03 19:34:43.921964 [NOTICE] switch_core_io.c:1202 Activating write
resampler

2016-08-03 19:34:43.961958 [WARNING] switch_core_codec.c:721 Codec SPEEX
Exists but not at the desired implementation. 16000hz 36ms 1ch

2016-08-03 19:34:43.961958 [ERR] switch_core_media.c:3021 Can't load codec?

 

The VBR is setting ptime to 36, 77, etc, varies every call coming in, which
fails to find a match on speex implementation .

 

Both opus and speex16 calls come in to echo, depending on if the browser is
web-rtc capable to support fallback.

 

We send send into echo, they press 1 to transfer here

 

root at ip-10-0-0-69:/opt/freeswitch/conf/dialplan/default# cat
bbb_echo_test.xml

<include>

  <extension name="bbb_echo_test_direct">

    <condition field="${bbb_authorized}" expression="true"
break="on-false"/>

    <condition field="destination_number" expression="^9196$|^9196(\d{5})$">

      <action application="set" data="vbridge=$1"/>

      <action application="answer"/>

      <action application="bind_digit_action"
data="direct_from_echo,1,exec:execute_extension,${vbridge} XML default"/>

      <action application="sleep" data="1500"/>

      <action application="echo"/>

    </condition>

  </extension>

</include>

 

Then  they are transferred.

 

root at ip-10-0-0-69:/opt/freeswitch/conf/dialplan/default# cat
bbb_conference.xml

<include>

    <extension name="bbb_conferences">

      <condition field="${bbb_authorized}" expression="true"
break="on-false"/>

      <condition field="destination_number" expression="^(\d{5})$">

     <action application="set" data="jitterbuffer_msec=20:400"/> 

        <action application="answer"/>

        <action application="conference" data="$1 at cdquality"/>

      </condition>

    </extension>

</include>

 

 

So master from 7/06 currently after setting the jitterbuffer on speex call
changes the PTIME to some number that doesn't match.

 

Opus calls work fine.

 

If I  comment out the jitterbuffer in dialplan the calls work for both opus
and speex.

 

Any help on how to get  speex to stay fixed at 20ms like it had worked in
previous with the jitterbuffer setting.

 

Can we set jitter buffer defaults for opus another way?

 

Thanks for the help.

 

Regards,

Stephen

 

HostBBB - Online Learning Solutions  

207 Technology Group Inc.   1-888-229-9756  skype: Stephen_Dame

 

-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://lists.freeswitch.org/pipermail/freeswitch-users/attachments/20160803/027cf0f7/attachment.html