Improve Microphone Quality



I’m looking to use the GSC3510 (latest firmware version) as part of a conferencing system on a LAN using asterisk and the opus codec. A GSC3510 will be placed on a table where a meeting will take place, and there will be people listening in another room with headphones via another enable sip phone.

I’m finding at the moment, even using the Opus codec, the audio quality from the microphone is low. If I make a recording on the local phone, it sounds fine, its just the audio over SIP/RTP seems low quality.

Because this is on a dedicated LAN, I’m not worried about it being high bandwidth.

Here is the outgoing sip/sdp from the phone.

o=5001 8000 8000 IN IP4 
s=SIP Call 
c=IN IP4 
t=0 0 
m=audio 50040 RTP/AVP 107 101 
a=rtcp:50041 IN IP4 
a=rtpmap:107 opus/48000/2 
a=fmtp:107 useinbandfec=0; sprop-maxcapturerate=16000; stereo=0; sprop-stereo=0 
a=rtpmap:101 telephone-event/8000 
a=fmtp:101 0-15 

I’ve noticed the sprop-maxcapturerate is set to 16000 can this be improved? Of course, there could be a problem somewhere else in the chain, but I thought I would start here.

Any help you can give to debug this issue would be appreciated. If this works out, we will be buying 10 of these units.


To be clear, are you indicating that the audio quality is poor or that the audio volume is “low”? I am not used to people as describing the audio quality as low. As the recording is done at the local phone and not at the SIP level, but taking the analog signal from the mic and converting into an audio file, the comparison may apples to oranges.


Hello, thanks for the reply.

Sorry for the confusion, I meant the audio quality is poor. The Audio level (loudness) is fine.

I’d argue the comparison is not completely void. It is my understanding that codecs like Opus allow for full bandwidth audio, and I was merely stating that I am not currently getting that based on what the unit is capable of achieving when recording.


I understand, but as OPUS is not used in the recording of the audio, then what you hear when playing back an audio encoded file as compared to a codec that has certain compression artifacts associated to it, is not a fair comparison as I suspect that no matter the codec, the direct playback of the audio file will always be better.

However, the point is taken in that the audio is not as good sounding as you would expect. The change in the sampling rate would possibly improve this. I indicate possibly, as much depends on the device delivering the sound as well as the hearing ability of the person listening. Empirically, and using objective measurements, the tonal frequency range would be expanded and many would perceive the audio qualify as better.

I can only say that as far as I am aware, there is no way to alter the sampling frequency or bandwidth used. I have suspicions that while the codec is capable of accommodating numerous variables to deliver a variety of hearing experiences, I have often wondered if the variables are limited given the purpose of the devices (telephony) and knowing that destinations will also include those that can’t handle the frequency range that OPUS can offer. There are a few providers that reportedly offer OPUS, but most don’t. I also suspect that many of the delivering devices may not have quality speakers that can handle the range. I wonder if they don’t limit the variables in order to force the devices to deliver a tonal frequency range that is considered to be more attuned to what the telephone network, phone can handle and what most users of same expect.

I don’t know that my opinion is worth anything and may be completely wrong, but other IP-phone systems that I have seen also impose limits.

You could submit a ticket to GS and see. It may also be interesting to see if you called in the reverse direction, what the other end offers in the codec.



Thank you for your feedback.

Have you test it / compare it with G722? From our testings, using the OPUS codec, the audio quality is acceptable and also our test result G722 have better quality compares to the rest of the codecs. You can give it a try to use G722.

Of course, there is always a room to improve and we are welcoming all the feedback from the customers if you still think the audio quality need to be improved.

Thank you,


Opus is better then G.722 as parameters show. So if this is reverse in GS then codec implementation is bad.
Of course you can have as good as quality mic/speaker used.


Marcin - I got curious about it and apparently that depends on the implementation. Opus is defined in RFC 6716 and it appears that the implementations I have seen in the IP telephony world all seem to follow the LP layer to include GS -

The Opus codec is a real-time interactive audio codec designed to meet the requirements described in [REQUIREMENTS]. It is composed of a layer based on Linear Prediction (LP) [LPC] and a layer based on the Modified Discrete Cosine Transform (MDCT) [MDCT]. The main idea behind using two layers is as follows: in speech, linear prediction techniques (such as Code-Excited Linear Prediction, or CELP) code low frequencies more efficiently than transform (e.g., MDCT) domain techniques, while the situation is reversed for music and higher speech frequencies. Thus, a codec with both layers available can operate over a wider range than either one alone and can achieve better quality by combining them than by using either one individually.

As we have seen the traces indicate 16Khz for Opus and we know G722 is 48khz, then the chart does indicate a higher MOS for g722.

I was not aware of this until I started looking as I was pretty much of the impression that it was the codec to be. However, I don’t use it as I am not aware of any providers that do (I don’t know them all) and in the GS world with the UCM, there is the MOH issue. But, more importantly, I prefer not to introduce various codecs as I do not need clients starting to compare the audio delivered by one source versus another and then wondering why all are not the same. So, I choose to use the more standard g711 throughout and avoid the issues that I might otherwise introduce. Just my preference.


I have one ITSP that add Opus also G.722 is used almost everywhere.
G.722 is much better then PCM so we recommend 722, Opus not as you point it is not fully supported.
From RFC
o 8-12 kbit/s for NB speech,
o 16-20 kbit/s for WB speech,
o 28-40 kbit/s for FB speech,

As we can see FB speech is not supported that’s why it is worse then G.722, also it could work VBR but it is not implemented i guess.

With this there is no point to use OPUS, 711 is better then (stable) and 722 have better quality.