audio-stream | Stream audio through Node.js backend to web frontend | Runtime Evironment library

This answer would solve a lot of my problems but relies on wc -l to tally the number of audio channels from the output of ffprobe.

How do I use ffmpeg to merge all audio streams (in a video file) into one audio channel?

I'm using a Windows batch file, so I need another way of accomplishing the following in CMD:

This job sets 2s segments for the video and audio streams. The video total duration is 134s, so I would expect about 67 segments. However we see in the MPD manifest that there are 45 video segments, and 54 audio segments (for each audio track).

Is this the expected behavior? Our player does buffer more than 2s at once.

Why is there a different number of video and audio segments?

Job Config

I am trying to use module.exports() to create a new module in my NW.js application.

I have two files that I am using:

Index.js

I have an html canvas that was created in p5, and I would like to add an audio track to it so that I can stream it with a webrtc connection. I currently can stream the visuals but not the audio.

I am adding the audio stream to my canvas as follows:

I am using a robot to hold conversations with volunteers. I am using python3 and Microsoft's Speech SDK to transcribe the volunteers responses. Both the recording and the transcription is done using the Speech SDK and I have not been able to find a way how to access and save the recorded audio file.

Minimal code example:

I am am using the AxWindowsMediaPlayer control to build a small Windows web radio developed in C#.

This works out well. My StatusChange event handler extracts the name of the current radio station:

I am using MediaInfo.dll with Wrapper-Class to check Video-Files for Audio-Codecs.

Can someone tell me how I can check COUNT of the Audio-Streams of the File?

Basicly, I'm trying to build and play audio data from bytes, that comes from WS sockets.

Detailed:

I have simple WS server written in Django-Channels, that on connect returns me splitted audio file in blob object with 6144 bytes of each chunk. Next, I want to decode this blob data and turn it into sound:

This is very early stages therefore no code, but some architecture questions maybe.

Im looking into trying to create a walkie-talkie functionality from ex a desktop-application or android application can send its audio to a server and that server then distributes the stream of audio to all clients.

My issue is that we are talking about both WIFI and LTE/4G network so has to work over the internet, and in theory should be possible to push audio from 1 to 1000 clients ( or select clients )

A small delay from speaker to its distributed isnt a big problem since its only one way communication ( not a like a phone with two way communication ).

Alot of questions arise here, mostly about size and speed :

1. primary thing im considering if i need to talk out to 1000 clients ( as in many many clients ) at the same time, i assume those 1000 clients all need a few sockets that is connected to the server, and therefore probably have to split it over more than one server to handle such kind of load ? ( i dont know ).

2. signalling part - would it be possible at all to have a signalling service handling that many clients ? ( assume the signalling needs also a constant connection to the server to be able to react when theres a audio-stream coming out, as it has to be fairly quick to react when someone speaks )

3. the protocols i have looked into on an overall perspective are SIP for signalling and RTP over TCP for transport, alternatively looked briefly at XMPP and fun-XMPP for inspiration, and in theory i can see it work on a small scale but my brain breaks when i try to imagine it on a large scale.

Core architecture was ex having a server handling SIP for signalling and keeping track of clients ( if SIP is fast enough/real-time enough which i am in doubt of ) - and then a transport server that the client would connect to with its audio-stream and then that server would channel that data out to all clients through the connection made when SIP recognizing someone is 'calling' them.

Side-question to this would be that i would like to have Go as back-end to manage this routing of data to all clients as a kind of media streaming server - but not sure if its fast enough ?

Maybe i am am totally wrong with the approach and protocols - maybe a way better approach then also interested in what you would suggest instead.