Deepgram
Deepgram is a voice AI platform offering speech-to-text, text-to-speech and voice agent APIs. It is mainly used to build transcription, call analytics and real-time voice agents.
#36 in Video & Audio ToolsTranscription SoftwareFree plan
Who it's for
Best for
Developers building real-time voice agents and call transcription
Businesses transcribing high volumes of noisy phone calls
Not for
- Teams whose main need is top-quality text to speech
Features
Speech-to-text APIs
Streaming and batch speech-to-text, including Flux STT models for English and multilingual audio.
Flux text to speech
Flux TTS voices are available over batch requests or WebSocket streaming for spoken replies.
Voice Agent API
A Voice Agent API combines listening, LLM orchestration, function calling and spoken responses.
Audio intelligence add-ons
Add-ons such as redaction, keyterm prompting and smart formatting are priced per minute of audio.
Free starting credit
Usage-based accounts start with a free $200 credit, with prepaid Growth plans from $4,000 a year.
What users say
Pros
- Noisy real-world calls with regional UK accents transcribe more accurately than with competing providers.
- Speaker diarization is accurate and fast on long multi-speaker meetings.
- Latency and pricing hold up for high-volume API transcription.
- End-of-thought triggers for voice-agent turn-taking are handled wonderfully.
- Remote dictation through the API is a big step up from Apple's built-in dictation.
Cons
- Text to speech trails other providers in quality.
- A speaker joining late in a long call, or two similar voices, can confuse diarization.
- Swedish transcripts come out rougher than ElevenLabs on names and compound words.
- Spoken British postcodes are recognised less often than with several rival models.
- Turn detection did not beat open-source models in one team's tests.
- Startup program credits expire after about six months.
A selection of the review sources we used:
Summarized from public user reviews and community discussions. View details
Global Traffic Rank
Top 500k
Alternatives
Rank#36 in category#9 in category#13 in category#14 in category
Free planYesYesYesYes
ProAccurate noisy-call transcripts, strong diarization, low latencyAccurate, fast transcripts and a generous free tierCuts typing on long recordings, handles accents, quick call notesGenerous free plan, accurate summaries, video replay, Ask Fathom
Better forTranscribing files in a browser with no code and free daily transcriptsReady-made meeting summaries and action items with no codingRecording video meetings with instant summaries and CRM sync
Weaker atReal-time streaming transcription for live voice appsBuilding custom voice apps and call analytics on speech APIsTranscribing high volumes of noisy phone calls