AssemblyAI · août 2026
11 mises à jour · août 2026
L'archive Spyingbee contient 11 pages suivies pour AssemblyAI, enregistrées pour la première fois en août 2026. Elles ne portent pas de date de publication : la date indiquée est celle de notre première collecte. Le thème dominant : feature updates (7). À retenir : « AssemblyAI adds real-time LLM Gateway for streaming audio analysis » (27 août 2026).
Parcourir d'autres mois de AssemblyAI
AssemblyAI introduced a new feature enabling real-time analysis of streaming audio transcripts via its LLM Gateway API. The update allows developers to integrate 25+ models (Claude, GPT, Gemini) into …
AssemblyAI introduced a hosted Qwen3.5 4B model optimized for voice rewrite tasks in its LLM Gateway, delivering 612ms average response times and 94% lower costs than GPT-4.1. The model is purpose-bui…

AssemblyAI now allows users to preserve filler words like 'uh' and 'um' in transcripts by setting `disfluencies` to `true` in the transcription config. By default, these are removed. The change is ref…
AssemblyAI introduced dedicated endpoints for US and EU data residency alongside its default global endpoint. The new endpoints (`sync.us.assemblyai.com` and `sync.eu.assemblyai.com`) route audio and …
AssemblyAI revised its streaming WebSocket protocol documentation to introduce new server-to-client messages like `SpeechStarted` (Universal-3.5 Pro only) and refined `Turn` message behavior, includin…
AssemblyAI’s LLM Gateway now supports 26 models, including new additions like Opus 4.7, Opus 4.8, Sonnet 5, and multiple GPT-5 variants, alongside existing models. Pricing varies by model, with GPT-5 …
AssemblyAI launched Universal 3.5 Pro Realtime, a next-generation streaming speech-to-text model, and integrated it with LiveKit’s Agents framework. The integration enables real-time transcription, tu…
AssemblyAI’s Voice Agent API now supports explicit language steering via the `language_codes` parameter, allowing developers to pin transcription to specific languages (e.g., Spanish-only) for higher …
AssemblyAI introduced Voice Focus in its voice agent API to isolate caller voices and suppress background noise before transcription. The feature supports near-field and far-field models, with adjusta…
AssemblyAI introduced inline session configuration for voice agents using `session.update` over WebSocket, allowing real-time agent behavior adjustments without server-side storage. The update support…
AssemblyAI’s Voice Agent API now supports a free-form transcription prompt and a key terms list to improve speech-to-text accuracy for domain-specific terms. The prompt sets the call context (e.g., ph…
Suivez AssemblyAI en pilote automatique