What is Speechmatics?
Speechmatics is a speech-to-text and voice AI API that transcribes audio in real time or in batches across 55+ languages for developers. The company is based in Cambridge, UK, and its customers include Adobe, Ubisoft, LiveKit, Vapi and AI Media. In an independent Pipecat benchmark from August 2026 it recorded a 1.07% pooled word error rate, the lowest of 12 services tested.
Speechmatics is known for handling accents, dialects and speakers who switch languages mid-sentence. Developers can choose between general, high-accuracy and medical models for recorded audio, and use a real-time model with built-in turn detection for voice agents. Add-ons cover translation, summaries, sentiment, chapters and topics, and there is also a text-to-speech service. It can run in Speechmatics' cloud, on-premise or on-device, which matters for regulated industries. Typical uses are captioning, call center analytics, meeting tools, voice assistants and clinical documentation.
Main Features
- Transcribe recorded audio files in batches through an API.
- Stream real-time transcription with turn detection for voice agents.
- Handle 55+ languages, including speakers who switch languages mid-sentence.
- Separate speakers with diarization and add word-level timestamps.
- Use a medical model tuned for clinical terminology.
- Add translation, summaries, sentiment, chapters and topics as extras.
- Deploy in the cloud, on-premise or on-device.
How to Get Started with Speechmatics
- Create a free account at speechmatics.com to receive $100 in credits.
- Generate an API key in the portal.
- Upload a test file in the portal or call the API with an SDK.
- Choose the model and options, then move to real-time streaming if needed.
Speechmatics Pricing
New accounts get $100 in free credits with no card required. Pre-recorded transcription starts at $0.12 per hour with the Melia 1 model, Standard costs $0.23 per hour and Enhanced costs $0.38 per hour, with add-ons such as translation at $0.49 per hour. Enterprise plans with volume discounts and dedicated support are custom.
Who Should Use It?
- Developers building voice agents and assistants.
- Media and captioning companies transcribing live and recorded content.
- Contact centers analyzing calls across many languages.
- Healthcare software teams handling clinical speech.
- Regulated businesses that need on-premise speech recognition.
Speechmatics Alternatives
- Gladia: a similar multilingual speech API with real-time support
- Rev: adds human transcription for maximum accuracy
- Amberscript: a ready-made transcription app instead of an API
Frequently Asked Questions
Is Speechmatics free?
Yes, to start. New accounts get $100 in free credits with no card required, then transcription is billed per hour from $0.12.
How many languages does Speechmatics support?
Speechmatics supports more than 55 languages. It can also handle speakers who switch between languages in the same conversation.
Can Speechmatics run on-premise?
Yes. Besides its cloud API, Speechmatics offers on-premise and on-device deployment for privacy-sensitive workloads.