Skip to main content

Documentation Index

Fetch the complete documentation index at: https://docs.gradium.ai/llms.txt

Use this file to discover all available pages before exploring further.

Deepgram integrations often already separate provider code into “send audio, receive transcript” or “send text, receive audio” adapters. With that structure in place, moving to Gradium is mostly an endpoint, auth-header, and field-name swap. Use Gradium’s STT endpoints when you are replacing Deepgram Listen, and Gradium’s TTS endpoints when you are replacing Deepgram Speak.

​
Endpoint swap

FlowDeepgramGradium
Pre-recorded STTPOST https://api.deepgram.com/v1/listenPOST https://api.gradium.ai/api/post/speech/asr
Streaming STTwss://api.deepgram.com/v1/listenwss://api.gradium.ai/api/speech/asr
One-shot TTSPOST https://api.deepgram.com/v1/speakPOST https://api.gradium.ai/api/post/speech/tts
Streaming TTSwss://api.deepgram.com/v1/speakwss://api.gradium.ai/api/speech/tts

​
Speech-to-text POST

For pre-recorded audio, keep sending the audio bytes in the request body and switch the URL to Gradium. Gradium streams newline-delimited JSON back; collect text messages to build the transcript. 
curl -L -X POST "https://api.deepgram.com/v1/listen?model=nova-3" \
  -H "Authorization: Token $DEEPGRAM_API_KEY" \
  -H "Content-Type: audio/wav" \
  --data-binary @input.wav

​
Speech-to-text WebSocket

Deepgram Listen streaming accepts audio over the WebSocket connection. Gradium does the same, but direct WebSocket audio frames are JSON messages with base64 audio.
Gradium messages
{"type":"setup","model_name":"default","input_format":"pcm"}
{"type":"audio","audio":"base64_encoded_audio"}
{"type":"end_of_stream"}
Deepgram Listen conceptGradium STT message
WebSocket connect to /v1/listenConnect to wss://api.gradium.ai/api/speech/asr
Audio stream{"type":"audio","audio":"base64_encoded_audio"}
Transcript result{"type":"text","text":"..."}
Close/finalize stream{"type":"end_of_stream"}
Set input_format to match the audio you send: pcm, wav, opus, ulaw_8000, alaw_8000, or another supported Gradium input format.

​
Text-to-speech POST

Deepgram Speak uses the model query parameter for voice/model selection and sends text in the JSON body. Gradium uses voice_id in the JSON body and returns raw audio bytes when only_audio is true. 
curl -L -X POST "https://api.deepgram.com/v1/speak?model=aura-2-thalia-en" \
  -H "Authorization: Token $DEEPGRAM_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{"text": "Hello from Deepgram."}' \
  > output.wav
Deepgram conceptGradium field
model query parameter for Speakvoice_id plus optional model_name
texttext
Output encoding optionsoutput_format
Authorization: Token ...x-api-key

​
Text-to-speech WebSocket

For streaming TTS, replace Deepgram Speak messages with Gradium’s setup, text, and end_of_stream messages:
Deepgram Speak messageGradium TTS message
{"type":"Speak","text":"..."}{"type":"text","text":"..."} after setup
{"type":"Flush"}Send more text, or finish with end_of_stream for a simple utterance
{"type":"Close"}{"type":"end_of_stream"}
Binary/audio response frames{"type":"audio","audio":"base64..."}
Gradium messages
{"type":"setup","voice_id":"YTpq7expH9539ERJ","model_name":"default","output_format":"wav"}
{"type":"text","text":"Hello from Gradium."}
{"type":"end_of_stream"}

​
Checklist

  • Replace the Deepgram URL with the matching Gradium endpoint.
  • Change auth to x-api-key.
  • For STT, set Content-Type or input_format to match your audio.
  • For TTS, pass a Gradium voice_id.
  • For direct WebSocket STT, send audio as base64 inside Gradium audio messages.
  • Keep your transcript collector or audio playback path.

​
Next steps

Gradium STT REST guide

Complete-file transcription with streamed NDJSON responses.

Gradium STT WebSocket guide

Real-time audio streaming, VAD messages, and flush.

Gradium TTS REST guide

One-shot text-to-speech with raw audio responses.

Gradium TTS WebSocket guide

Streaming text-to-speech over WebSocket.