Speechmatics
Send audio via API or SDK, select model or deployment mode, and receive low-latency transcripts with speaker diarization and language detection. Speechmatics supports cloud, on-prem, and on-device deployments, offers starter credits for evaluation, and documents per-hour model rates and enterprise custom contracts for large volumes or private models.
New accounts receive $100 in credit with no card required; Pro is pay-as-you-go with usage rates and volume discounts; Enterprise is custom-priced with unlimited scale and dedicated support.Is Speechmatics right for you?
Good for
- Product teams building voice agents, live captioning, or regulated transcription workflows who need multilingual, low-latency speech-to-text with flexible deployment modes.
- Covers 55+ languages, including support for speakers switching languages mid-sentence (code-switching).
- Real-time low latency suitable for live captioning and voice agents, reporting sub-second latency for STT.
Keep in mind
- Model-training discount requires opting in, which grants permission for audio and transcripts to be used to improve models.
- Free starter access has limited real-time concurrency (2 concurrent sessions), which can block high-concurrency evaluation.
- Buyers who need publicly posted SLAs or uptime guarantees on the pricing pages; confirm operational SLAs and concurrency limits with Speechmatics for production use.
Pricing and plan limits
The pricing pages list per-hour model rates (for batch and real-time) and bolt-on feature prices; full enterprise pricing is custom and requires contacting sales.
More about capabilities and limits
Documented strengths
Covers 55+ languages, including support for speakers switching languages mid-sentence (code-switching).
Real-time low latency suitable for live captioning and voice agents, reporting sub-second latency for STT.
Multiple deployment modes: on-device, on-premises, container, virtual appliance, and multi-region cloud options for privacy-sensitive use.
Limitations to consider
Model-training discount requires opting in, which grants permission for audio and transcripts to be used to improve models.
Free starter access has limited real-time concurrency (2 concurrent sessions), which can block high-concurrency evaluation.
Choosing a plan
Use the $100 starter credit to validate accuracy, concurrency, and deployment mode, then review Pro rate cards and enterprise options with Speechmatics to confirm capacity and SLA needs.
Based on the linked product documentation.
Available on
Api.
Official sources
Alternatives to Speechmatics
Choose around the work you need to do.
AssemblyAI
Use AssemblyAI when you need a single cloud Voice AI API that supports realtime and pre-recorded STT plus structured speech-understanding tools and a no-code playground for integration testing.
