speech-to-speech

LIVE PULSE

3.9 Anthropic CEO Amodei calls for slower AI development and shared safety rules — 11 src2.1 Agility Robotics unveils Digit 5 humanoid for warehouses and factories — 2 src2.0 Apple ships rebuilt Siri with Google Gemini, but not in the EU — 2 src1.7 Siri AI in macOS 27 Golden Gate: FAQ, Germany availability, privacy questions — 2 src1.4 Sam Altman says OpenAI will not go public in 2026, citing AI safety concerns — 5 src1.1 OpenAI contractors review real ChatGPT conversations to rate responses, report says — 2 src1.1 Anthropic data retention policy prompts firms to limit Claude use for sensitive work — 1 src1.1 Study examines issue bias in LLMs used as writing assistants before Swedish 2026 election — 1 src1.1 Study Audits Misalignment in Multi-Modal World Models — 1 src1.1 Retrieval-Grounded Reasoning Approach Proposed for Universal Multimodal Embeddings — 1 src3.9 Anthropic CEO Amodei calls for slower AI development and shared safety rules — 11 src2.1 Agility Robotics unveils Digit 5 humanoid for warehouses and factories — 2 src2.0 Apple ships rebuilt Siri with Google Gemini, but not in the EU — 2 src1.7 Siri AI in macOS 27 Golden Gate: FAQ, Germany availability, privacy questions — 2 src1.4 Sam Altman says OpenAI will not go public in 2026, citing AI safety concerns — 5 src1.1 OpenAI contractors review real ChatGPT conversations to rate responses, report says — 2 src1.1 Anthropic data retention policy prompts firms to limit Claude use for sensitive work — 1 src1.1 Study examines issue bias in LLMs used as writing assistants before Swedish 2026 election — 1 src1.1 Study Audits Misalignment in Multi-Modal World Models — 1 src1.1 Retrieval-Grounded Reasoning Approach Proposed for Universal Multimodal Embeddings — 1 src

topic2 events

papersTODAY 04:00 UTC

Causal Analysis and Mitigation of Spurious Speech Onsets in Full-Duplex Speech LLMs

A new arXiv paper examines why full-duplex speech models such as Moshi and its PersonaPlex derivative sometimes start talking when the user has gone quiet, including on digital-zero input. The authors trace these unwanted onsets to specific causes in the generation process and propose ways to reduce them. The work targets improving turn-taking reliability in speech-to-speech systems.

speech/safety/full-duplex/research

papersSEP 10 04:00 UTC

RelayS2S: Dual-Path Speculative Generation for Real-Time Speech-to-Speech Dialogue

A new arXiv paper proposes RelayS2S, a dual-path speculative generation method for real-time spoken dialogue systems. It addresses the trade-off between latency and response quality, since end-to-end speech-to-speech models can respond instantly and manage turn-taking, backchanneling, and interruptions, but tend to produce semantically weaker replies. The method aims to combine immediate responsiveness with improved response content.

speech-to-speech/real-time/dialogue/latency