Speechmatics is a speech-to-text and text-to-speech API built for developers who need automatic transcription across many languages inside their own products — call centers, media platforms, meeting tools and voice AI agents. It runs on a usage-based model: a genuinely free tier covering 3,000 minutes of speech-to-text and 1 million characters of text-to-speech per month, a Pro tier billed from roughly $0.129/hour with volume discounts, and custom Enterprise pricing for unlimited scale.
Who builds it
Speechmatics is built by Speechmatics Ltd, a UK-based speech-AI company. Developers integrate its REST/streaming API into their applications to convert audio to text (or text to speech) in real time or via batch file jobs, choosing between its Enhanced, Standard and Melia 1 models depending on accuracy, latency and language needs.
Core features
- Automatic speech recognition covering 56+ languages, with real-time (streaming) and batch (file-based) transcription modes
- Text-to-speech covering roughly 20 hours of audio per 1 million characters processed
- Three selectable models — Enhanced, Standard and Melia 1 — trading off accuracy, latency and cost, with Melia 1 adding 69 language pairs for translation and multilingual transcription
- Concurrent real-time session limits that scale by tier: 2 on Free, 50 on Pro, unlimited on Enterprise
- File-job throughput up to 10 jobs/second on Pro, with no rate limits on Enterprise
- Enterprise options for private/on-premises deployment, custom model and voice development, and a dedicated Customer Success Manager
Pricing
- Free: $0, no credit card required — 3,000 minutes (50 hours) of speech-to-text per month, 1 million characters (~20 hours) of text-to-speech per month, 2 concurrent real-time sessions
- Pro: from roughly $0.129/hour with a 20% discount option, keeping the same 3,000 free minutes and 1M free TTS characters monthly before pay-as-you-go rates apply; usage is capped at 6,000 hours/month; volume discounts apply automatically above 500 hours/month
- Enterprise: custom pricing — unlimited scale and concurrent sessions, no rate limits, private deployment options, custom models
- Check speechmatics.com/pricing for current per-hour rates, since usage-based API pricing can change
Who it's for
Individual developers and small projects prototyping speech features can run comfortably on the free 3,000-minutes/month tier. Growing products with real-time transcription needs — call centers, meeting apps, media monitoring — fit the Pro tier once they exceed the free allowance. Organizations needing guaranteed uptime, private/on-premises deployment, custom voice models, or usage well beyond 6,000 hours/month should talk to Speechmatics about Enterprise pricing.