Welcome to pyannote Discussions!

[hbredin]

👋 Welcome!

We’re using Discussions as a place to connect with other members of pyannote community. We hope that you:

• Ask questions you’re wondering about.
• Share ideas.
• Engage with other community members.
• Welcome others and are open-minded. Remember that this is a community we build together 💪.

To get started, comment below with an introduction of yourself and tell us about what you do with pyannote!

[ajay960singh]

Hello guys,

We have an audio stream running on an RTMP server and we are using pre-trained models to do inference on the data. We tried to use librtmp library in Python to read the stream dn extract the audio data. We used the following code to get the stream data-

import librtmp

# Create a connection
conn = librtmp.RTMP("rtmp://your.server.net/app/playpath", live=True)
# Attempt to connect
conn.connect()
# Get a file-like object to access to the stream
stream = conn.create_stream()
# Read 1024 bytes of data
data = stream.read(1024)

This is the result when I print data:

b'FLV\x01\x05\x00\x00\x00\t\x00\x00\x00\x00\x12\x00\x01M\x00\x00\x00\x00\x00\x00\x00\x02\x00\nonMetaData\x03\x00\x08duration\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x08filesize\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x05width\x00@\x94\x00\x00\x00\x00\x00\x00\x00\x06height\x00@\x86\x80\x00\x00\x00\x00\x00\x00\x0cvideocodecid\x02\x00\x04avc1\x00\rvideodatarate\x00@i\x00\x00\x00\x00\x00\x00\x00\tframerate\x00@>\x00\x00\x00\x00\x00\x00\x00\x0caudiocodecid\x02\x00\x04mp4a\x00\raudiodatarate\x00@`\x00\x00\x00\x00\x00\x00\x00\x0faudiosamplerate\x00@\xe7p\x00\x00\x00\x00\x00\x00\x0faudiosamplesize\x00@0\x00\x00\x00\x00\x00\x00\x00\raudiochannels\x00@\x00\x00\x00\x00\x00\x00\x00\x00\x06stereo\x00?\xf0\x00\x00\x00\x00\x00\x00\x00\x07encoder\x02\x00\x1eRogue Amoeba RTMP Streamer 1.0\x00\x00\t\x00\x00\x01X\x08\x00\x00\x04\x00\x00\x00\x00\x00\x00\x00\xaf\x00\x11\x90\x00\x00\x00\x0f'

The data object contains encoded data with FLV tags and due to insufficient documentation , we have been unable to use data to convert to WAV format to input to pre-trained models. So is there a way to do this? Or are there other ways to run the pre-trained models directly from the RTMP stream?