AI Briefing#speech-to-text
The latest AI and developer news about #speech-to-text, with the original source and a short summary.
The latest AI and developer news about #speech-to-text, with the original source and a short summary.
Scribe v3 Live reduces realtime semantic errors by 43% and adds speaker diarization, while Scribe v3 improves batch accuracy in every tested language.
The guide explains the SRT file structure, how to edit timing errors, and when to choose SRT over WebVTT for video distribution.
The guide identifies audio input, speaker dynamics, and lexical issues as the primary causes of transcription errors and provides specific fixes for each.
The guide explains the four-step transcription pipeline and notes that post-call processing is more accurate and cost-effective than real-time streaming.
The character-level model outperforms all tested aligners across 29 languages and handles real-world noise with 57 ms average error.
A practical guide details how to build an AI voice assistant using Eleven v4 Turbo and Scribe v2 Realtime to achieve low-latency, high-accuracy interactions.
The release includes a LiveKit-based Voice Agent SDK and speech models priced from $0.0035 per minute for transcription and $13 per 1M characters for TTS.
ElevenLabs outlines a five-step workflow for podcast transcription, featuring Scribe v2 which supports 90+ languages, 32 speakers, and 10-hour files, with exports for Apple Podcasts and Spotify.
The update makes all of Cohere's latest model architectures compatible with llama.cpp and allows fine-tuning of Cohere Transcribe in Unsloth.
The guide details using Scribe v2 for 98% accurate transcription and caption generation, supporting up to 32 speakers and 1,000 key terms.
Achieving approximately 96% usage among direct foreign personnel, the app bridges communication gaps with shipbuilding-specific translation.
The new 8B and 30B models integrate a Decision aLoRA, Guardian aLoRA, and the fastest ASR model on the OpenASR Leaderboard while maintaining base-model VRAM usage.
The tool transcribes one hour of audio in approximately 13 seconds on a base M3 MacBook Air using the Neural Engine.
ElevenLabs Scribe v2 supports up to 1,000 keyterms per transcript to improve accuracy for names and jargon.
ElevenLabs Scribe v2 Realtime offers sub-150ms latency and 90+ language support as a production-grade alternative to the browser-dependent Web Speech API.
Scribe v2 Realtime achieves 150 ms latency for live captioning, while batch Scribe v2 supports up to 1,000 keyword terms for accurate post-meeting transcripts.
The guide highlights ElevenLabs' Scribe v2 and Scribe v2 Realtime models, which support diarization for up to 32 speakers and transcription in over 90 languages.
Scribe outputs structured token arrays with word, spacing, and audio_event types in a single API call, supporting up to 5 independent channels.
ElevenLabs' Scribe v2 model supports over 90 languages and speaker diarization for up to 32 participants, exceeding Zoom's native 19-language post-meeting limit.
ElevenLabs has released a Speech to Text integration tutorial for Python developers, with Scribe v2 Realtime offering latency under 150ms.
ElevenLabs has officially launched Scribe v2 Medical, optimized for clinical audio, significantly improving medical ASR accuracy.
ElevenLabs released Scribe v2 Medical, a speech recognition model specialized for medical audio, improving clinical audio WER by 18%.
Qwen has released Qwen3.8-LiveTranslate, a real-time interpretation model featuring speaker timbre preservation and source-translation synchronization.
xAI has launched the Grok Voice Transcribe 2.0 speech recognition model, doubling accuracy while maintaining the same price.