Experiment: ChatterBoxTTS

1 files changed, 38 insertions, 0 deletions

diff --git a/makima/src/main.rs b/makima/src/main.rs
index 9097ef6..2348b23 100644
--- a/makima/src/main.rs
+++ b/makima/src/main.rs
@@ -1,6 +1,44 @@
+use std::path::Path;
+use crate::tts::{save_wav, ChatterboxTTS};
+
+mod audio;
 mod listen;
+pub mod tts;
 
 fn main() -> Result<(), Box<dyn std::error::Error>> {
+    println!("Loading ChatterboxTTS...");
+    let mut tts = ChatterboxTTS::from_pretrained(None)?;
+    println!("Model loaded successfully!");
+
+    // // Voice cloning using existing audio file
+    // println!("Generating TTS with voice cloning...");
+    // let audio = tts.generate_tts_with_voice(
+    //     "Hello, this is a test of the voice cloning system.",
+    //     Path::new("audio.wav")
+    // )?;
+    //
+    // println!("Generated {} samples", audio.len());
+    // save_wav(&audio, Path::new("output.wav"))?;
+    // println!("Saved to output.wav");
+
+
+    // Load reference audio from mp3
+    println!("Loading reference audio...");
+    let reference = audio::to_16k_mono_from_path(Path::new("audio.mp3"))?;
+    let samples = &reference.samples;
+    let sample_rate = reference.sample_rate;
+
+    // Voice cloning using audio samples
+    println!("Generating TTS with voice cloning...");
+    let audio = tts.generate_tts_with_samples(
+        "Hello, this is a test of the voice cloning system [chuckles]. Repeat after me \" I am Steve Jobs!\"",
+        samples,
+        sample_rate,
+    )?;
+
+    println!("Generated {} samples", audio.len());
+    save_wav(&audio, Path::new("output.wav"))?;
+    println!("Saved to output.wav");
     let segments = listen::listen()?;
     println!("Captured {} diarized segments", segments.len());
     Ok(())