voice-training

Fine-tuning toolkit for Chatterbox TTS & Chatterbox TURBO models. Supports 23 languages with smart vocabulary extension. Features offline preprocessing, automatic VAD trimming, and voice cloning capabilities. Train custom TTS models with your own dataset in LJSpeech and file-based format.

面向 MTF、crossdresser 与性别多元成年人的中文 Agent Skill，基于生如夏花知识库并加入医学安全护栏。

Best Voice Training Software 2026: Visualize Formant, Weight, and Pitch with Rust Precision

Public showcase for Voiceform, an on-device voice mirror for real-time voice feedback

Voice analysis for training your speaking voice on Linux/Windows — pitch, formants, resonance and voice quality, tracked across sessions. Built on Praat.

micdrp: gets you out of the shower and onto the stage.

Privacy-first, 100% in-browser Voice Impression & Pitch Analysis web app powered by Transformers.js, WebGPU, and ONNX. 瀏覽器端純本地聲學與語音印象分析工具。

Standalone native macOS application tracking the main applio repo, with additional pretrained model (RefineGAN 44.1 kHz) for improved high-quality voice conversion. See README_MACOS.md

Free preview pack of high-fidelity 48kHz clinical AI voice data. Optimized for STT/TTS benchmarks and medical transcription pipelines.

Rust-based voice trainer to help visualize formant, weight, and pitch while training safely.

Production-grade fine-tuning & LoRA toolkit for Chatterbox-Flash zero-shot TTS models. Combines parallel block diffusion and FlashInfer acceleration with smart placeholder vocabulary extension supporting languages. Features offline feature preprocessing, Silero VAD silence trimming, zero-padding leakage prevention, and fast voice cloning for custom

Voice Training Tool

`inaSpeechSegmenter` web UI.

Fine-tuning toolkit for Maya1, a 3B-parameter open-source TTS model by Maya Research. Supports full fine-tuning with SNAC neural codec, natural language voice design, and inline emotion tags. Features YAML-based config, offline preprocessing, and automatic audio sample generation at every checkpoint.

Open-source, transfemme-focused voice-training practice companion.

Persistent voice training with Qwen3-TTS on Apple Silicon—fully local and reference-free at inference.

This repository serves as the official open-source evaluation hub for a premium, high-fidelity Conversational Female Monologue Dataset. This data addresses the critical shortage of natural human velocity, spontaneous breath placement, and unscripted vocal cadence in traditional training corpora.

My voice training exercises!

Free, 100% local real-time voice training tool for trans & non-binary voices: pitch, tones, recording, analysis, EN/FR

Privacy-first Android pitch coach for voice practice, guitar tuning, and local recording review.