Deepgram
A speech AI provider offering transcription models such as Nova-3 and text-to-speech voices such as Aura-2 through a REST API.
In more detail
Its pre-recorded transcription returns word-level timestamps, and its speech API limits each request to 2,000 characters of text. Its transcripts make a useful independent check on generated speech.
Related terms
- speech to text — Converting spoken audio into text, often with a timestamp for each word.
- text to speech — Generating spoken audio from text with a voice model.
Taught in
- Images, Audio And Speech — AI Engineering: LLM Apps, RAG & Agents
References
- Deepgram: Pre-recorded audio (nova-3 word timestamps)
- Deepgram: Text to speech REST API (2,000 character input limit for Aura-2)