Soniox STT v6 adds agent text input and echo filtering for voice agents
Soniox released Soniox STT v6, an updated speech-to-text API that transcribes both caller and agent speech in real time. The model uses agent text input to recognize short replies and filter out echo from the agent’s voice, improving accuracy in cascaded voice-agent systems. It ships as stt-rt-v6 (streaming) and stt-async-v6 (file-based), supporting 60+ languages via Soniox’s cloud API only.…
Key points
- Soniox STT v6 transcribes both caller and agent speech in real time via **stt-rt-v6** (streaming) and **stt-async-v6** (file-based)
- Agent text input enables echo filtering by feeding the agent’s spoken words to the model, improving accuracy in voice-agent calls
- Pricing starts at **$0.12/hour for streaming** and **$0.10/hour for async**, with agent text input billed at **$4.00 per 1M tokens**
Soniox claims v6 is significantly more accurate than its predecessor but has not published benchmark results yet. The company cites v5’s 1.34% semantic WER on Pipecat, trailing Azure’s 1.21% mean WER but with faster latency. v5 models will be deprecated December 15, 2026, with automatic routing to v6. The update also improves handling of numbers, commands, accents, and language switching. Soniox positions v6 as the first model to filter echo using agent text, though AssemblyAI’s Universal-3.5 Pro Realtime also supports agent context in streaming STT.
Soniox STT v6 Adds Agent Text Input and Echo Filtering for Voice Agents
MarkTechPost · 3 November 2026
Loading the full article…
This text was published by MarkTechPost and written by Michal Sutter. It is reproduced here with attribution so you can read it in full; the rights remain with the publisher. Read it at the source ↗
The headline, key points and digest above were generated by Digest AI's editorial model from the linked sources. Automated summaries can contain errors: the sources are the record. Spotted a mistake? Tell us. Published by Martin K., who runs Digest AI and handles corrections.
More in Agents & Tools
All →- Convoso makes no-code Voice AI Agent generally available · 1 src
- Scikit-LLM tutorial shows LLM summarization in scikit-learn pipelines using Groq free tier · 1 src
- Are Skills from Claude Code Becoming a Shared Asset for AI? · 1 src
- Anthropic adds message threads to Claude Platform for faster agent responses · 3 src
- OpenAI’s Codex flags AI work logs for deviations from user requests · 1 src
Comments
via GitHub Discussions