MediaStream

[Centurius123]

Hey,

I try to send a MediaStream via socket.io-stream, but I dont know how.

Is it possible to do this?

[PMLMonteiro]

Hi Centurius123

It is definitely possible.
I am currently doing it using the MediaRecorderStream API.
Just grab the stream from navigator.getUserMedia and record it.
You can then pipe the generated stream into an Socket.io-stream stream.

Hope it helps.

[vin-ni]

Have you found a solution to this?

[amber-cd]

Almost a year later, hope I'm not necroposting, but I'm running out of ideas here... has anyone else found a more detailed solution to this? How exactly is someone supposed to use media-recorder-stream to do this? I feel like I'm missing something.

To expand on my use case, I need to stream live audio from a microphone to my service on the back end, which then, using credentials stored in the service, will send that audio stream to an audio transcription service and emit the transcription as it updates. My specific situation here:

1. I don't want to send to the transcription service directly because it involves a private authentication key and I don't trust the front end with it.
2. I want this transcription to appear in live time, so it's not sufficient to wait for the MediaRecorderStream to stop and then send a stream with that blob data; I need something which will receive data every time the MediaRecorderStream does, within one stream which is piped to the transcription service.

Which leaves me kind of lost... what am I supposed to do from media-recorder-stream that will be the live equivalent of createBlobReadStream from the audio blob?

(I can create a new issue if need be, or move this to Stack Overflow as is appropriate; this is just the closest to a solution I've found when frantically googling.)

[vin-ni]

Hi @amber-workxpress ,
take a look at this: https://github.com/vin-ni/Google-Cloud-Speech-Node-Socket-Playground
This might be what you're looking for :)

[amber-cd]

@vin-ni Oh my god, you are a LIFESAVER. Thank you so much; that was exactly what I needed. I had to modify the code a bit to work with my React and Flux-based setup, but this gave me exactly the piece I was missing.

[enahsor]

Just in case anyone has a similar issue, WebRTC allows you to do this with its RTCPeerConnection API.

[vin-ni]

Just in case anyone has a similar issue, WebRTC allows you to do this with its RTCPeerConnection API.

could you point to an example?

[enahsor]

Sure @vin-ni
https://gist.github.com/enahsor/bacbd4a770e080ae224b7b57894b2224

[enahsor]

Sure @vin-ni
https://gist.github.com/enahsor/bacbd4a770e080ae224b7b57894b2224

@vin-ni did it work for you?
@Centurius123 maybe you can try this?

[vin-ni]

Haven't tried it yet. But will once I need something like that the next time and give you feedback!

[Krishnakumarskr]

Hi @amber-workxpress ,
take a look at this: https://github.com/vin-ni/Google-Cloud-Speech-Node-Socket-Playground
This might be what you're looking for :)

Wow! You are a lifesaver! Thank you so much!. This repository cleared all my doubts.
I have been stuck on many stages while implementing this. Also, I couldn't find helpful answers on StackOverflow. But you did it.
Thanks again!

Some unanswered SO questions:
https://stackoverflow.com/questions/56051454/mediastream-pipe-to-nodejs-socket-io-stream-to-google-speech-api-and-stream-back