Unverified60% confidenceFactExact time
Codec-based approaches (such as VALL-E and SoundStorm) encode speech into discrete tokens and then generate output using a language model approach, achieving better zero-shot voice cloning results.
2
Sources
60%
Confidence
Long-term
Relevance
8/10/2026
First Seen
Sources
Airy语音内容创作工具：免费快速简单的AI配音方案
Related Claims
UnverifiedNative audio architectures perform self-attention computation directly on mel spectrograms or quantized audio encodings (such as the discrete audio tokens produced by EnCodec), thereby capturing prosody, emotion, and speaker identity information absent from text representations.76% similarVerifiedZero-shot synthesis learns general rules of the voice space by pre-training on massive multi-speaker datasets, thereby generalizing new speaker characteristics from short audio clips.75% similarUnverifiedVALL-E, released by Microsoft in 2023, models speech generation as a language model problem, requiring only 3 seconds of reference audio to achieve zero-shot cloning, and uses the EnCodec neural audio codec.75% similarUnverifiedVoxtral uses a 3-billion-parameter text model as its backbone and converts audio into continuous embeddings via an audio encoder that are fed directly into the decoder, which differs from Whisper's approach of feeding into the decoder via cross-attention.73% similarUnverifiedVoice cloning technologies such as Microsoft's VALL-E and open-source projects Bark and RVC can generate highly realistic voice imitations from just a few seconds of voice samples.73% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/726730API
curl https://kongchang.com/api/v1/knowledge/claims/726730MCP
get_claim(id=726730)