AssemblyAI / python-websocket-client

Websocket Client Demo

Install Python dependencies (Mac OSX only)

This will work on Ubuntu as well, you'll just need to use apt to install portaudio (below) instead of brew.

brew install portaudio

If you don't have Homebrew, you can install it with /usr/bin/ruby -e "$(curl -fsSL https://raw.githubusercontent.com/Homebrew/install/master/install)"

pip install -r requirements.txt

Stream Audio File

This script will stream an audio file to the AssemblyAI WebSocket API. Transcripts will show up in the terminal as the file is uploaded.

python stream_file.py websocket.assemblyai.com <your api token> /path/to/file.wav

Audio files must be in the following format:

• Single channel (mono)
• 8000 sampling rate
• 16-bit Signed Integer PCM encoding (wav file)

If you are unsure, we recommend using sox to inspect the audio file by running: soxi /path/to/audio.wav. The output should look something like this:

Input File     : '/Users/me/Downloads/myaudiofile.wav'
Channels       : 1
Sample Rate    : 8000
Precision      : 16-bit
Duration       : 00:00:32.94 = 263499 samples ~ 2470.3 CDDA sectors
File Size      : 527k
Bit Rate       : 128k
Sample Encoding: 16-bit Signed Integer PCM

Stream from Microphone

This script will stream audio from your machine's microphone to the AssemblyAI WebSocket API. Transcripts will show up in the terminal as you speak.

python stream_mic.py websocket.assemblyai.com <your api token>