Verified75% confidenceFactExact time
The new end-to-end voice model is trained directly at the audio level, without first converting speech to text, and can directly capture acoustic features and emotions.
3
Sources
75%
Confidence
Long-term
Relevance
7/11/2026
First Seen
Sources
实测OpenAI新语音模型：可打断、同声传译、情绪捕捉全解析
bilibili神烦老狗7/9/2026
Related Claims
VerifiedEnd-to-end voice models take audio directly as input and output, performing understanding and generation at the audio token level while preserving acoustic features such as intonation, speaking rate, sighs, and laughter.84% similarUnverifiedThe new generation voice AI model takes audio directly as modal input, skipping the transcription step to achieve more natural emotional perception and intonation variation.83% similarUnverifiedThe workflow offers two voice generation modes: a lip-sync mode (which freezes the user's voice into the audio latent space) and a model-generated speech mode.81% similarVerifiedThe new generation of voice interaction has shifted to an end-to-end multimodal model architecture that understands and generates directly at the speech signal level, compressing latency and preserving paralinguistic information.81% similarUnverifiedCurrent flagship voice products generally adopt a joint training approach combining large-scale pretrained language models with dedicated speech decoders.81% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/485586API
curl https://kongchang.com/api/v1/knowledge/claims/485586MCP
get_claim(id=485586)