Ce que Speechmatics a publié en 2026

Real-time speech-to-text that scales with you

Speechmatics unveiled a 20% accuracy improvement in its real-time speech-to-text engine, now supporting 55+ languages with under 1-second latency. The update outperforms competitors like Microsoft, As…

On-device speech-to-text, in 10 questions

Speechmatics introduced On-Device speech-to-text, running entirely on end-user machines with no audio sent to servers, achieving within 5% of cloud accuracy and beating Whisper checkpoints by 12-16% i…

Scale more easily with Speechmatics

Speechmatics will add credit packs and monthly/annual subscription options from October 1, 2026, alongside pay-as-you-go. Discounted pricing for high-volume users (500+ hours/month) will shift from pa…

Introducing Agent STT: built for the words that cost you the call

Speechmatics introduced Agent STT, a speech-to-text API optimized for voice agents, powered by its new Linden model. Agent STT targets errors that break agent workflows (e.g., alphanumeric strings, ac…

Speechmatics Agent STT SDK adds external turn detection and Python client

Speechmatics updated its Agent STT service with a new Python SDK and configurable turn detection. The SDK now supports external turn closure via `finalize()` calls, allowing applications to control wh…

The future of AI is nameless: Tresic selects Speechmatics to power the speech layer of its conversation intelligence platform

Tresic selected Speechmatics to power the speech-to-text layer of its Intelligence Cloud platform after evaluating against Google and Deepgram. Speechmatics runs within Tresic’s infrastructure, ensuri…

Melia 1 leads speech-to-text code-switching in Arabic, Mandarin and Tamil

Speechmatics' Melia 1 model now leads in code-switching speech-to-text accuracy for Arabic, Mandarin, and Tamil, with mixed error rates as low as 15.1% (Arabic/English) and switch point F1 scores up t…

Speechmatics & LiveKit Inference target the accuracy gap that breaks voice agents in production

Speechmatics launched Linden, a speech-to-text model optimized for voice agents, now accessible through LiveKit Inference. Developers can switch to Linden in under a minute via UI or code, with billin…

Gone in 20 seconds: Melia now outpaces AssemblyAI, ElevenLabs and Soniox on Artificial Analysis

Speechmatics introduced Melia, a speech-to-text model achieving a 206x speed factor on Artificial Analysis' leaderboard, outperforming competitors like AssemblyAI and ElevenLabs. Melia processes an ho…

AI Voice Agents vs Chatbots: Which Delivers Better Customer Outcomes?

Speechmatics published a comparative analysis asserting AI voice agents deliver superior customer outcomes over chatbots in urgent, complex, or emotionally charged interactions due to faster resolutio…

Pricing change detected for Speechmatics

Pricing updated for Speechmatics: - Free: allowance changed from 56+ languages, 2 concurrent real-time sessions, $100 in credit to 55+ languages, 2 concurrent real-time sessions, $100 in credit - Pro:…

Pricing updated for Speechmatics: - Free: allowance changed from 56+ languages; 2 concurrent real-time sessions; $100 in credit to 56+ languages, 2 concurrent real-time sessions, $100 in credit - Pro:

The most accurate speech-to-text, now one Zap away

Speechmatics is now available on Zapier, offering two actions—synchronous Get Transcript from Audio and asynchronous Submit Audio (Webhook)—to connect its speech-to-text engine to 8,000+ apps without …

Pricing updated for Speechmatics: - Enterprise: allowance changed from 56+ languages; No rate limits; Custom models; Highest concurrency to 56+ languages; No rate limits; Custom models; Highest concur

Pricing updated for Speechmatics: - Free: allowance changed from 56+ languages, $100 in credit, 2 concurrent real-time sessions, multi-region cloud options, low-latency (ideal for voice agents) to 56+

Pricing updated for Speechmatics: - Free: new promotion — $100 in credit to get started, no card required - Free: allowance changed from 3,000 minutes (50 hours) per mon

Community is at the core of every voice agent

Speechmatics highlighted its community as a hub for voice-agent development, featuring real-time transcription, speaker diarization, and integrations with LiveKit, Pipecat, and Vapi. The page showcase…

Pricing updated for Speechmatics: - Free: promotion ended (was: $100 in credit to get started, no card required) - Free: allowance changed from 56+ languages, 2 concurre

Speechmatics launches Medical Model for real-time clinical transcription

Speechmatics introduced a domain-specific Medical Model for real-time clinical transcription, achieving 93% accuracy and 50% fewer keyword errors than competitors. The model supports seven languages (…

Pricing updated for Speechmatics: - Pro: promotion changed — $100 in credit to get started, add a card when you're ready; Sign up now to lock in lowest price - Removed: