qwen_tts 0.1.1

Qwen3-TTS text-to-speech model implementation for Candle
Documentation 
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15

//! 12Hz Audio Tokenizer (V2).
//!
//! This tokenizer converts between audio waveforms and discrete codes at 12.5Hz.
//! It consists of:
//! - Encoder (based on Mimi): Audio → Codes
//! - Decoder: Codes → Audio
//!
//! The decoder uses:
//! - Split Residual Vector Quantizer for dequantization
//! - ConvNeXt blocks for upsampling
//! - Transformer for sequence modeling
//! - Snake activation-based vocoder

pub mod v2;
pub mod wrapper;