Voice Cloning & Conversion¶
Audio
Open Source¶
-
CosyVoice 3 / Qwen-Audio-3.0-TTS Discovered by AI
2026
A technical retrospective of the CosyVoice lineage (from CosyVoice to CosyVoice 3 and Qwen-Audio-3.0-TTS), detailing a scalable multilingual zero-shot text-to-speech system that separates speech planning via an autoregressive language model from acoustic rendering via flow matching.
flow-matchingttsvoice-cloning -
Higgs Audio Discovered by AI
2026
Higgs Audio (Higgs TTS 3) is a 4-billion-parameter multilingual text-to-speech foundation model designed for natural, conversational voice interactions. It supports zero-shot voice cloning from a short reference clip across over 100 languages, with fine-grained inline control over emotion, style, prosody, and sound effects.
conversational-aimultilingualttsvoice-cloning -
MOSS-TTS Discovered by AI
2026
MOSS-TTS is an open-source speech and sound generation foundation model family. It features high-fidelity, zero-shot voice cloning, controllable long-form synthesis, and multilingual/code-switched speech generation, using a custom 12Hz audio tokenizer and autoregressive modeling.
audio-generationmultilingualttsvoice-cloningGitHub · arXiv · Hugging Face
-
Qwen3-TTS Discovered by AI
2026
Qwen3-TTS is an open-source multilingual text-to-speech series trained on over 5 million hours of speech data across 10 languages. It supports zero-shot voice cloning (CustomVoice) and natural language-based voice design, delivering highly expressive and human-like speech generation.
multilingualspeech-synthesisttsvoice-cloningGitHub · arXiv · Hugging Face
-
ZONOS2 Discovered by AI
2026
ZONOS2 is an open-weight text-to-speech model based on a sparse Mixture of Experts (MoE) architecture with 8 billion total parameters. It is trained on over 6 million hours of multilingual speech and supports high-fidelity, naturalistic zero-shot voice cloning, as well as granular control over emotion, speed, and tone.
moemultilingualttsvoice-cloningGitHub · arXiv · Hugging Face
-
CosyVoice 3 Discovered by AI
2025
An advanced multilingual zero-shot text-to-speech and voice cloning model that scales up training data and introduces post-training techniques to handle in-the-wild speech generation with high naturalness and speaker similarity.
ttsvoice-cloningzero-shot -
VibeVoice Discovered by AI
2025
An expressive, long-form conversational speech synthesis and voice cloning model by Microsoft that utilizes next-token diffusion and a highly compressed continuous speech tokenizer.
diffusionttsvoice-cloningGitHub · arXiv · Hugging Face
-
VoiceCraft-Dub Discovered by AI
2025
An automated video dubbing framework that synthesizes high-quality speech from text and facial cues. It extends neural codec language models by incorporating video features, ensuring that the cloned voice is temporally synchronized and expressively aligned with facial movements.
dubbingmultimodalvoice-cloning -
VoiceCraft-X Discovered by AI
2025
An autoregressive neural codec language model that unifies multilingual speech editing and zero-shot text-to-speech synthesis across 11 languages. It leverages the Qwen3 large language model and a novel token reordering mechanism to perform high-fidelity voice cloning and editing.
codec-lmspeech-editingvoice-cloning -
Chatterbox Discovered by AI
2024
An open-source text-to-speech and zero-shot voice cloning model by Resemble AI, featuring emotion control, real-time generation, and built-in watermarking.
ttsvoice-cloningzero-shot -
CosyVoice 2 Discovered by AI
2024
An advanced multilingual streaming text-to-speech and voice cloning model that integrates instruction-following and zero-shot capabilities. It achieves human-parity naturalness and ultra-low latency by combining a large language model with a conditional flow matching decoder.
streamingttsvoice-cloning -
DreamVoice Discovered by AI
2024
A text-guided voice conversion framework that allows users to specify desired voice timbres and styles through natural language prompts instead of reference audio clips.
diffusiontext-guidedvoice-conversion -
FunAudioLLM (CosyVoice) Discovered by AI
2024
FunAudioLLM is a family of speech foundation models featuring CosyVoice for natural speech generation. CosyVoice supports zero-shot multilingual voice cloning, cross-lingual voice cloning, and fine-grained control over timbre, speaking style, and emotion.
multilingualttsvoice-cloning -
MaskGCT Discovered by AI
2024
MaskGCT is a zero-shot text-to-speech and voice cloning model based on a masked generative codec transformer, eliminating the need for explicit alignment or duration predictors. It achieves highly natural and faithful voice replication using only a 3-second audio prompt.
audio-generationttsvoice-cloningzero-shotGitHub · arXiv · Hugging Face
-
Seed-TTS Discovered by AI
2024
Seed-TTS is a family of large-scale autoregressive text-to-speech and voice cloning models developed by ByteDance. It generates speech that is virtually indistinguishable from human speech, supporting zero-shot voice cloning, in-context learning, and high controllability over emotion and style.
speech-synthesisvoice-cloningzero-shot -
Seed-VC Discovered by AI
2024
A zero-shot voice conversion and singing voice conversion framework based on diffusion transformers, featuring real-time support and an external timbre shifter to prevent timbre leakage.
diffusionsinging-voicevoice-conversion -
Seed-VC Discovered by AI
2024
An open-source zero-shot voice and singing voice conversion framework featuring real-time streaming inference, timbre transfer, and built-in voice anonymization capabilities.
singing-voicevoice-cloningvoice-conversionzero-shot -
VoiceCraft Discovered by AI
2024
A token infilling neural codec language model that achieves state-of-the-art performance on speech editing and zero-shot text-to-speech synthesis in the wild.
codec-lmspeech-editingvoice-cloning -
VoiceStudio Discovered by AI
2024
A fully local, open-source alternative to ElevenLabs for voice cloning, voice design, video dubbing, and text-to-speech across hundreds of languages without requiring external API keys.
local-firstttsvoice-cloning -
XTTS Discovered by AI
2024
XTTS is a massively multilingual zero-shot text-to-speech and voice cloning model that supports over 17 languages. Building upon the Tortoise architecture, it enables high-quality voice replication and cross-lingual voice cloning from a short audio prompt of only a few seconds.
multilingualvoice-cloningzero-shot -
Amphion Discovered by AI
2023
An open-source audio, music, and speech generation toolkit that hosts state-of-the-art models like MaskGCT for zero-shot text-to-speech and voice cloning without requiring explicit duration or alignment prediction.
audio-generationttsvoice-cloning -
Applio Discovered by AI
2023
An advanced, highly optimized open-source voice conversion tool based on Retrieval-based Voice Conversion (RVC). It features a clean, user-friendly interface, active community support, and various performance enhancements for high-quality voice cloning and transformation.
rvcvoice-cloningvoice-conversion -
Bark Discovered by AI
2023
A highly popular transformer-based text-to-audio model capable of generating highly realistic speech, music, and background noise, supporting zero-shot voice cloning via audio prompts.
transformerttsvoice-cloning -
OpenVoice Discovered by AI
2023
A versatile instant voice cloning approach that replicates a speaker's voice from a short audio clip and enables granular control over styles, including emotion, accent, rhythm, and intonation.
ttsvoice-cloningzero-shot -
Realtime Voice Changer (w-okada) Discovered by AI
2023
A real-time AI voice changer and voice conversion application that supports multiple model backends including RVC, MMVC, and Beatrice, widely used by VTubers and creators.
real-timevoice-changervoice-conversion -
RVC (Retrieval-based Voice Conversion)
2023
A popular open-source voice-conversion toolkit that retargets a source voice to a trained timbre, widely used for character voice and dubbing experiments.
open-sourcevoice-conversion -
Voicebox: Text-Guided Multilingual Universal Speech Generation at Scale Discovered by AI
2023
Meta's non-autoregressive flow-matching generative model trained on 50,000 hours of speech, performing zero-shot voice cloning, speech denoising, and cross-lingual voice style transfer.
flow-matchingspeech-generationvoice-cloningzero-shot -
Coqui TTS Discovered by AI
2021
A widely used deep learning toolkit for Text-to-Speech and voice cloning, featuring the powerful XTTS-v2 model which allows cloning voices in multiple languages using just a 6-second audio clip.
ttsvoice-cloningxtts
Proprietary / Industry Tools¶
-
Altered Studio Discovered by AI
2022
A professional AI voice content creation platform and voice changer that allows users to clone voices, morph speech into curated AI voices, and perform advanced audio editing for game and media production.
voice-changervoice-cloningvoice-conversion -
ElevenLabs Voice Design & Cloning Discovered by AI
2022
An industry-leading proprietary platform for high-fidelity AI voice generation, instant voice cloning, and professional voice replication used in game and media production.
dubbingproprietaryttsvoice-cloning -
Lovo AI Discovered by AI
2019
An AI voice generator and text-to-speech platform featuring the Genny editor, which combines realistic voice cloning, multi-language synthesis, and video editing capabilities.
ttsvoice-cloningvoiceover
