Top updates from 2025
November 2025
- Launchspeechmatics.com
Best TTS APIs in 2026: ElevenLabs, Google, AWS & 9 More Compared
Speechmatics introduced the first bilingual Voice AI model supporting Arabic and English, marking a global first in the industry. This innovation expands their speech-to-text capabilities to multiling…
October 2025
- speechmatics.com
Speechmatics Medical Model launches in Spanish
Speechmatics expanded its Medical Model to support Spanish, Dutch, Finnish, and French, achieving 7.3% WER in Spanish—20% better than competitors like AssemblyAI and Whisper. The models include real-t…
- speechmatics.com
Why we built our low-latency Text-to-Speech
Speechmatics introduced a new streaming Text-to-Speech (TTS) API delivering under 150ms latency while maintaining natural voice quality, addressing gaps in real-time voice agents. The product supports…
- speechmatics.com
Vapi and Speechmatics: Build agents that understand every voice
Speechmatics announced native integration with Vapi, enabling developers to build voice AI agents with Speechmatics' industry-leading speech recognition as the input layer. The integration supports re…
September 2025
- speechmatics.com
Introducing real-time, speaker-aware Voice Agents with LiveKit + Speechmatics
Speechmatics and LiveKit announced a partnership enabling LiveKit’s 100,000 developers to access Speechmatics’ real-time, speaker-aware speech recognition and diarization. The integration adds speaker…
August 2025
- speechmatics.com
Knowing who said what: the importance of Speaker Diarization for analytics and conversations in Voice AI
Speechmatics details its self-supervised learning approach to speaker diarization, achieving 48% fewer speaker identification errors and 38% fewer speaker change mistakes at 1-second latency compared …
July 2025
- speechmatics.com
Speechmatics launches world’s first bilingual voice AI models for Southeast Asia
Speechmatics introduced the first bilingual voice AI models for Southeast Asia—Mandarin-English, Malay-English, and Tamil-English—designed to handle natural code-switching in real-world conversations.…
June 2025
- speechmatics.com
The scaling challenge voice AI can’t ignore
Speechmatics details how real-time concurrency underpins its ability to process over 500 years of human conversation monthly, supporting sessions up to 48 hours and beyond. The company argues that con…
May 2025
- speechmatics.com
The return of on-premise: Why enterprise AI's head is no longer in the cloud
Speechmatics introduced Flow On-Premise, enabling enterprises to deploy its Conversational AI API locally for total data sovereignty, eliminating third-party cloud risks. The solution runs on Kubernet…
- speechmatics.com
From fjords to phonetics: Mastering the complexities of Norwegian speech-to-text
Speechmatics reduced transcription errors in Norwegian by 40%, eliminating 1,400 mistakes per hour of audio. The improvement stems from model scaling, diverse dialect training data, and language-speci…
Track Speechmatics on autopilot
- · Weekly AI brief: narrative summary of what shipped, every Monday 9 AM
- · Email or Slack alerts, or chat with the archive in your dashboard
- · Add Speechmatics + up to 2 more competitors free, no credit card