My question is about the huge difference in the audio quality between the human voice and a musical instrument - for example, a piano - when using video chat apps as Skype, FaceTime, Facebook messenger, etc. Even in the best scenario, when the connection is decent and both the picture and the audio quality are good enough to communicate, the musical sound is transmitted with huge distortion, the effects ranging from sounding "underwater" to extremely percussive, not to mention the latency. Is there anything that can be done to resolve this issue?
3 Answers
Many of the codecs (audio compression algorithms like e.g. GSM or G.729) used in internet communication are intended for speech, not music. They may be based on human vocal tract model (nasal cavity, oral cavity, pharynx, larynx, trachea, velum, tongue). This specialization allows high compression ratio (= low bandwidth use) and relatively good quality for speech but may give terrible effects when processing other sounds.
Other problem that may contribute is volume difference between voice and (presumably) quiet background music. This background may be treated as noise and filtered out or distorted by denoise filter or partially cut out by voice activity detection algorithm that can stop transmission if signal is below certain level or send so called "comfort noise" instead - again, to minimize bandwidth used.
-
Many thanks for this info. That certainly makes sense. Does this means that a totally new platform has to be developed in order to cater to one being able to play musical sounds online (for example when teaching a piano lesson online) or there is/can be a plug in that can be added to a video chat app?– R. HayJul 13, 2017 at 21:45
-
I don't know any service/operator I can recommend, but you could also try to use two softphones paired in peer to peer mode (direct IP calling). One inconvenience of it is that you may have to forward ports, at least on one side. There are softphones that can send/receive L16/44100 or L16/48000 (uncompressed samples) and allow to disable any audio processing. Major issue for music might be delay, this paper might be interesting: mm.aueb.gr/publications/2015-MusiNet-IISA.pdf– TMSZJul 14, 2017 at 12:32
If you would like something good for both voice and music, try the OPUS codec. Not all VoIP clients has support for it, but nowadays is quickly becoming the most important codec in VoIP due to its superior quality vs bandwidth needs.
-
Many thanks for your response. I've looked at it and it sounds like this may well be exactly what I need. However, as a non-technical person (I'm a musician), I can't see if OPUS can be easily used together with another platform like Skype, Google Hangout or FaceTime, etc. (as a plugin) or a lot more work has to be done to incorporate it into a video chat app?– R. HayJul 13, 2017 at 21:50
-
Actually Skype were using an OPUS variant many years ago (with failback to narrowband codec when the network quality is wrong). Otherwise, I don't think that you can touch the codec's of any of these apps. The good news is the more and more VoIP clients are starting to use OPUS.– IstvanJul 17, 2017 at 18:58
Zoom seems to be the only platform for which you have an option for not compressing the sound. Definitely the only way to treat musical signal. Vital for music lessons for instance.