Verified65% confidenceFactExact time
A typical EnCodec configuration uses 8 layers of RVQ with a codebook size of 1024 per layer, producing 75 frames per second at a 24kHz sampling rate, compressing 1 second of audio into 600 tokens.
3
Sources
65%
Confidence
Long-term
Relevance
8/5/2026
First Seen
Sources
H3语音模型全精度权重发布：表现力惊艳但长句声音漂移待解
Related Claims
UnverifiedMeta的EnCodec在24kHz音频上仅需1.5kbps即可实现高质量重建，约为传统MP3的十分之一79% similarUnverifiedThe standard Bluetooth SBC codec has an upper bit rate limit of approximately 328 kbps.72% similarUnverifiedBy default, EmbeddingGemma 2 uses 280 tokens per image, 140 tokens per video frame, approximately 25 tokens per second of audio, and has a context window of 8,192 tokens.70% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/687639API
curl https://kongchang.com/api/v1/knowledge/claims/687639MCP
get_claim(id=687639)