DeepTranscript provides state-of-the-art speech-to-text accuracy in both english and french languages. In this repository you will see how simple it is to integrate to DeepTranscript High Availability API.

Batch mode

You have wav, mp3 or flac audio recordings and you want an accurate transcription of what has being said ? batch mode is all you need !

CURL example

$ curl https://app.deeptranscript.com/api/transcriptions/ \
  --request POST \
  --header "Authorization: Bearer [API_TOKEN]" \
  --header "Content-Type: application/json" \
  --data '{"recording":"[PUBLIC_URL_TO_AUDIO_RECORDING]","recordingFormat":"mp3","callbackUrl":"[URL_TO_CALL_ONCE_TRANSCRIPTION_READY]","language":"en"}'

See ./examples/batch-mode and API documentation for more information

Streaming Mode

Deeptranscript API provides two streaming options: HTTP streaming and WebSocket streaming

With HTTP streaming, you can pipe up to 60s of raw audio data to a standard HTTP request. Once input stream ends you get transcription results in no time.

cURL example

$ getAudioStreamSomehow | curl \
  -H "Authorization: Bearer [API_TOKEN]" \
  -X POST \
  --data-binary @- \
  https://stream.deeptranscript.com/\?language\=fr\&sampleRate\=8000\&format\=s16le\&localizeWords\=1

see ./examples/streaming/http and API documentation for more information

With WebSocket streaming, you can process input streams of any size and get intermediate results indicating if user is speaking and what is being said in almost realtime.

see ./examples/streaming/websocket and API documentation for more information

Before you begin

• Make sure you signed up and get your 20h evaluation welcome bonus
• Retrieve your API Token from members configuration
• Check requirements and dependencies

Code examples requirements

• python 3.6+
• nodejs v10+

Setup (debian or ubuntu)

$ sudo apt install ffmpeg curl git python3-venv unzip flac
$ git clone [email protected]:DeepTranscript/deeptranscript-demo.git deeptranscript-demo
$ cd deeptranscript-demo

# setup python virtualenv
$ python3 -m venv ./venv
$ source ./venv/bin/activate
$ (venv) python --version  # make sure it is 3.6+
$ (venv) pip install -U pip
$ (venv) pip install flask yt-dlp requests

# setup node dependencies
$ (venv) npm install

# setup ngrok (batch-mode only)
$ curl https://bin.equinox.io/c/4VmDzA7iaHb/ngrok-stable-linux-amd64.zip --output ./ngrok.zip \
  && unzip ./ngrok.zip \
  && rm ./ngrok.zip \
  && chmod +x ngrok

Troubleshooting

If anything goes wrong you will find informations in DeepTranscript Console

Bad request

Ensure the protocol and port are matching (wss and 443)

$ websocat ws://stream.deeptranscript.com:443/
websocat: WebSocketError: Received unexpected status code (400 Bad Request)
websocat: error running