Struct TtsEngine 

pub struct TtsEngine { /* private fields */ }

Main TTS engine for speech synthesis from GGUF models.

Implementations

impl TtsEngine

pub async fn new<P: AsRef<Path>>(gguf_path: P) -> Result<Self>

Create a new TTS engine from a GGUF model file.

Arguments

• gguf_path - Path to the Q4 GGUF model file

Returns

A TTS engine ready for synthesis, or an error if model loading fails.

Example

use voxtral_tts::TtsEngine;
let tts = TtsEngine::new("models/voxtral-tts-q4.gguf").await?;

pub async fn with_options<P: AsRef<Path>>( gguf_path: P, tokenizer_path: Option<P>, voices_dir: Option<P>, ) -> Result<Self>

Create a new TTS engine with custom options.

Arguments

• gguf_path - Path to the Q4 GGUF model file
• tokenizer_path - Optional path to tokenizer JSON (auto-discovered if None)
• voices_dir - Optional path to voices directory (defaults to models/voxtral-tts/voice_embedding)

pub fn synthesize( &mut self, text: &str, voice: Option<&str>, ) -> Result<AudioBuffer>

Synthesize speech from text using the default voice (casual_female).

Arguments

• text - The text to synthesize
• voice - Optional voice name (defaults to “casual_female”)

Returns

An audio buffer containing the synthesized speech at 24kHz.

pub fn synthesize_with_options( &mut self, text: &str, voice: Option<&str>, speed: f32, gain: f32, language: Option<&str>, ) -> Result<AudioBuffer>

Synthesize speech with custom options.

Arguments

• text - The text to synthesize
• voice - Optional voice name (defaults to “casual_female”)
• speed - Playback speed multiplier (0.5 to 3.0, where 1.0 is normal)
• gain - Volume gain multiplier (0.1 to 2.0, where 1.0 is normal)
• language - Optional language code (e.g., “en”, “fr”, “de”)

Returns

An audio buffer containing the synthesized speech at 24kHz.

Speed Behavior

Speed adjustment uses resampling which changes both tempo and pitch. Higher speeds result in higher pitch (chipmunk effect), lower speeds result in lower pitch.

Example

let mut tts = TtsEngine::new("models/voxtral-tts-q4.gguf").await?;
let audio = tts.synthesize_with_options(
    "Hello world!",
    None,      // voice: None = default "casual_female"
    2.0,       // speed: 2.0 = twice as fast
    0.8,       // gain: 0.8 = 20% quieter
    Some("en") // language
)?;

pub fn list_voices(&self) -> Result<Vec<String>>

List available voice presets in the voices directory.

Returns

A vector of voice names, or an error if the directory doesn’t exist.

pub fn set_max_frames(&mut self, max_frames: usize)

Set the maximum number of audio frames to generate.

Default is 2000 frames. Higher values allow longer audio generation.

pub fn set_euler_steps(&mut self, steps: usize)

Set the number of Euler ODE steps for flow matching (quality vs speed tradeoff).

• 3 steps: Real-time performance
• 4 steps: Balanced (default)
• 8 steps: Higher quality

pub fn save_wav<P: AsRef<Path>>( &self, path: P, audio: &AudioBuffer, ) -> Result<()>

Save audio buffer to a WAV file.

Arguments

• path - Output file path
• audio - Audio buffer to save