Deepgram Speech-to-Text
Speech-to-text API supporting 50+ languages with streaming and batch transcription, speaker identification and smart formatting, alongside companion text-to-speech and voice-agent APIs.
Find alternatives to Deepgram Speech-to-Text
What it does
- Speech to text
- Voice agent
Features 23
- Industry-tuned modelsSpecialized speech-to-text models tuned for vocabulary in domains such as healthcare, legal and finance.
- Custom modelsSpeech-to-text models trained on a customer's proprietary or novel datasets for edge cases.
- Real-time low latencyDelivers transcripts in under 300 milliseconds for voice agents and conversational AI.
- Multilingual supportA single speech-to-text API supporting 50+ languages.
- End-of-Turn DetectionDetects when a speaker has finished their turn, with configurable turn-taking, for use in voice agent pipelines.
- Keyterm PromptingLets requests supply key terms to improve recognition, for both monolingual and multilingual transcription.
- Smart FormatApplies additional formatting to transcripts so they are easier for people to read.
- DiarizationSeparates speakers in transcribed audio, using versioned diarization models.
- NumeralsConverts numbers in transcripts from written-out words to numerical format.
- Filler WordsTranscribes filler words, such as hesitations, instead of leaving them out of the transcript.
- RedactionOptionally redacts personal information from transcripts.
- MultichannelReturns a separate transcript for each audio channel when multichannel audio is submitted.
- Language DetectionIdentifies the dominant language spoken in the submitted audio.
- SearchFinds terms or phrases in audio by matching acoustic patterns, and returns where they occur.
- UtterancesGroups transcribed speech into utterances, helping the model handle speakers' spontaneous speech patterns.
- Profanity FilterMasks offensive language in transcripts using asterisks.
- DictationConverts spoken dictation commands into their corresponding punctuation marks in the transcript.
- MeasurementsAutomatically converts spoken measurement units into their abbreviations in the transcript.
- EndpointingDetects speech boundaries by monitoring incoming streaming audio.
- Interim ResultsProvides interim transcripts while monitoring streaming audio, before the final result is available.
- Utterance EndHelps detect the end of speech while transcribing live streaming audio.
- CallbackLets a request supply a callback URL to which the transcription is returned.
- Whisper CloudDeepgram-hosted version of the Whisper model, offered so users need not run their own.
How it compares
-
Speech to Text API
Deepgram Alternative: Better Accuracy, 55+ Languages, Lower Cost | Speechmatics
Official documentation · 19 Sep 2026 -
Sonic / Ink
Cartesia positions Ink/Sonic as Deepgram's alternative for speech-to-text and TTS in voice agents.
Cartesia | Deepgram alternative for speech to text and TTS
Company disclosure · 9 Oct 2026 -
AssemblyAI Pre-recorded Speech-to-Text API
Deepgram's own comparison page is titled as a head-to-head with AssemblyAI.
AssemblyAI vs Deepgram: Is Deepgram the Right Alternative?
Company disclosure · 14 Oct 2026 -
Amazon Transcribe
Deepgram's speech-to-text outshines Amazon Transcribe in both speed and accuracy
Official documentation · 13 Oct 2026 -
Cloud Speech-to-Text
Deepgram is 53% more accurate, nearly 40x faster, and 5x more affordable than Google STT.
Official documentation · 13 Oct 2026 -
Azure Speech in Foundry Tools
Deepgram is 30% more accurate, more than 25x faster, and over 4x more affordable than Microsoft.
Official documentation · 13 Oct 2026 -
OpenAI API
Deepgram is 36% more accurate, up to 5x faster, and has lower TCO than OpenAI Whisper.
Official documentation · 13 Oct 2026 -
Gladia Speech-to-Text API
Real customer calls, accents, overlapping speakers, languages switching mid-sentence. That’s where Gladia beats Deepgram.
Official documentation · 17 Oct 2026
Sources
- Pricing
-
Pay As You Go is metered; the only free offer is a one-off "Free $200 Credit", which is a trial credit, not a free tier.
- Description
-
Official documentation · 10 Oct 2026
Product page: "support 50+ languages".
- Constraints
-
Official documentation · 2 Oct 2026
Derived from the security facts on this page, from the same source.
- Security
-
Official documentation · 2 Oct 2026
Platform-wide Deepgram docs (data-privacy-compliance, your-data, model-improvement program). Requests join the Model Improvement Program by default; mip_opt_out=true excludes them and the your-data page says this gives zero data retention. GDPR described as 'ready' with no DPA page for EU stated, so gdpr-dpa not recorded (only an AU DPA is mentioned). apac = Australia-only endpoint. Voice Agent third-party LLM think step may process outside region. BAA available on request.
- Sold within
-
The pricing page offers "Pay As You Go" — "No minimums. No expiration." "No credit card required." — starting at "Free $200 Credit" then pay-as-you-go, with a "Sign Up Free" route; its Speech to Text table prices Nova-3 Monolingual streaming at $0.0048/min.
- Deployment
-
Official documentation · 18 Oct 2026
"Deepgram's cutting-edge voice AI is available to self-host on your own infrastructure, both in the cloud or on-premises." The Dedicated page (deepgram.com/dedicated) adds a single-tenant deployment managed by Deepgram.
- Interfaces
-
Official documentation · 14 Oct 2026
Page: 'This guide walks you through transcribing pre-recorded audio with the Deepgram API using cURL or one of Deepgram’s SDKs'.
- Modalities
-
Official documentation · 1 Oct 2026
Page: 'Convert speech-to-text with unmatched accuracy, ultra-low latency, and enterprise scalability.'
- Intended for
-
"Trusted by the world's top Enterprises and Startups"; solutions for "developers building voice agents, chatbots, or contact centers"
Also from Deepgram
-
Deepgram Flux TTS API service
Deepgram's text-to-speech API, marketed on low latency (responses from about 80ms) and on being deployable in the customer's own environment as well as on Deepgram's cloud.
-
Deepgram Voice Agent API API service
A single API for building real-time voice agents, combining Deepgram's speech-to-text and text-to-speech with LLM orchestration.
-
Deepgram Audio Intelligence API service
Language-understanding features layered on Deepgram's speech recognition — summarisation, topic and intent detection over conversational audio.
Open Deepgram in the directory
Something wrong here? Send a correction — quote this product id: deepgram-speech-to-text.