Ce que Speechmatics a publié en juin 2026

Speechmatics vs Google: Which Speech-to-Text API Delivers?

Speechmatics asserts its Ursa 2 model achieves 92% accuracy (G2 Spring 2026) versus Google’s 87%, with superior handling of noisy, accented audio and no cloud dependency. It also introduces Melia, a m…

Speechmatics vs Azure Speech: Which Speech-to-Text API Delivers?

Speechmatics introduced Ursa 2 (Standard & Enhanced) and Melia, a new multilingual model supporting 55+ languages with native code-switching, positioning itself as a specialist alternative to Microsof…

Speechmatics vs ElevenLabs Scribe: Which Speech-to-Text API Delivers?

Speechmatics positions itself as a superior speech-to-text (STT) API compared to ElevenLabs Scribe, emphasizing full real-time speaker diarization, true on-premises deployment, and stronger non-Englis…

Speechmatics vs OpenAI: Which Speech-to-Text API Delivers?

Speechmatics positions its Ursa 2 speech-to-text model as superior to OpenAI’s Whisper and gpt-4o-transcribe, emphasizing native real-time streaming with included speaker diarisation, on-premises/air-…

Speechmatics vs AWS: Which Speech-to-Text API Delivers?

Speechmatics positions itself as a specialist speech-to-text provider outperforming AWS in accuracy, accent handling, and deployment flexibility. It highlights a 4.8/5 G2 rating (vs AWS’s 3.9/5), sing…

Pricing change detected for Speechmatics

Pricing updated for Speechmatics: - Free: promotion ended (was: 8 hours free per month to try technology) - Free: allowance changed from 3,000 minutes (50 hours) per month | 1 million characters (~20 …

Speechmatics vs Soniox: Which Speech-to-Text API Delivers?

Speechmatics unveiled Ursa 2, its flagship STT model, positioning it as superior to Soniox across accuracy, scalability, and deployment flexibility. Ursa 2 supports 55+ languages, offers on-device and…

Speechmatics expands SDK support with new language integrations

Speechmatics released official SDKs for Python, JavaScript, .NET, and Rust to simplify integration of its speech-to-text and text-to-speech services. The SDKs support real-time streaming and batch pro…

Speechmatics launches GPU Speech-to-Text inference container

Speechmatics introduced a GPU-optimized inference container for its speech-to-text service, requiring Nvidia GPUs with 16GB+ memory, CUDA 7.5-9.0 support, and 24GB RAM. The container supports batch an…

Pricing updated for Speechmatics: - Free: new promotion — 8 hours free per month to try technology - Free: allowance changed from 3,000 minutes (50 hours) per month; 1 million characters (~20 hours) p

Speechmatics vs Whisper On-Device: how Adobe Premiere's speech engine got rebuilt

Speechmatics compressed its cloud-grade speech-to-text model via quantization to run efficiently on consumer laptops, outperforming Whisper on accuracy and speed while staying within Adobe’s resource …

A Practical Guide to Building Voice AI Applications With Real-Time Transcription in 2026

Speechmatics released a detailed guide on constructing production-ready voice AI applications using real-time transcription, emphasizing the critical role of transcription accuracy, latency, and syste…

Pricing updated for Speechmatics: - Free: allowance changed from 3,000 minutes (50 hours) per month + 1 million characters (~20 hours) per month to 3,000 minutes (50 hours) per month; 1 million charac

Speechmatics vs Gladia: Which Speech-to-Text API Delivers?

Speechmatics published a detailed comparison showing its Ursa 2 model outperforms Gladia in noise adaptation (94% vs 73%), speaker identification (84% vs 75%), and real-time diarization availability. …

How to Add Automatic Captions to Media Content Using a Speech-to-Text API

A blog post by Speechmatics explaining the process of integrating their speech-to-text API to generate automatic captions for media content.

How Modern Law Firms Can Use AI Transcription Without Compromising Client Data

Speechmatics argues that AI transcription can improve legal workflows without compromising client confidentiality by using secure platforms with transparent data handling, private cloud/on-premises op…

Pricing updated for Speechmatics: - Free: allowance changed from 2,400 minutes (40 hours) per month; 1 million characters (~20 hours) per month; 2 concurrent real-time sessions to 3,000 minutes (50 ho

The Complete Guide to AI Medical Transcription for Clinical Workflows in 2026

AI medical transcription is gaining traction in 2026 by reducing documentation time in clinical workflows, converting spoken encounters into structured records. Modern systems handle medical terminolo…

Introducing Melia, our new multilingual speech-to-text model

Speechmatics introduced Melia, a new multilingual speech-to-text model supporting code-switching across all 55+ languages in their catalog. It outperforms competitors like Deepgram, Microsoft, and Ass…

How to Integrate Multilingual Speech Recognition Into a Global Product

Speechmatics published a guide on integrating multilingual speech recognition, emphasizing real-world challenges like accents, dialects, code-switching, and language detection. It argues that broad la…

8 Common Medical Transcription Errors in Healthcare: How to Avoid Them

Article by Speechmatics outlining frequent transcription errors in healthcare and providing guidance to avoid them, positioning their speech recognition technology as a solution.

What Is an AI Voice Agent? Everything You Need to Know

Speechmatics publishes an educational article explaining AI voice agents, their functionality, and use cases. The content positions Speechmatics as an authority in speech recognition and AI-driven voi…

Real-Time vs. Batch Transcription: What's the Difference?

Speechmatics explains the technical differences between real-time and batch transcription, highlighting use cases and performance trade-offs for their speech-to-text technology.

Pricing updated for Speechmatics: - Free: promotion ended (was: 8 hours free per month to try technology; additional usage requires credit card) - Free: allowance changed from 2,400 minutes/month (rea

Speech-to-Text vs. Voice Recognition: What's the Difference?

An educational article explaining the distinction between speech-to-text and voice recognition technologies, likely aimed at clarifying Speechmatics' offerings in the voice AI space.

Pricing updated for Speechmatics: - Free: new promotion — 8 hours free per month to try technology; additional usage requires credit card - Free: allowance changed from 2,400 minutes/month, 1 million

Better than Whisper: how Adobe Premiere's on-device speech engine got rebuilt

Speechmatics compressed its cloud-grade speech-to-text model via quantization to run efficiently on consumer laptops, outperforming Whisper on accuracy and speed while staying within Adobe’s resource …

Pricing updated for Speechmatics: - Free: allowance changed from 480 minutes/month, 1M characters (~20hrs)/month, 2 concurrent real-time sessions to 2,400 minutes/month, 1 million characters (~20hrs)/

What Is Speech-to-Text AI and How Does It Work?

An educational article by Speechmatics explaining the fundamentals of speech-to-text AI technology, its mechanisms, and applications.

Pricing updated for Speechmatics: - Free: allowance changed from 480 minutes/month, 1M characters/month (~20hrs), 2 concurrent real-time sessions to 480 minutes/month, 1M characters (~20hrs)/month, 2