speech editing / zero-shot TTS
VoiceCraft is a token infilling neural codec language model for zero-shot speech editing and text-to-speech, cloning or editing an unseen voice with only a few seconds of reference audio.
Solves the same problem with a different architecture or at a different layer. Expect to rewrite the integration.
choose for few-shot voice cloning TTS with fast training and small data, if you don't need speech editing.
1 min voice data can also be used to train a good TTS model! (few shot voice cloning)
choose for multilingual tokenizer-free TTS with creative voice design, if you don't need speech editing.
VoxCPM2: Tokenizer-Free TTS for Multilingual Speech Generation, Creative Voice Design, and True-to-Life Cloning
choose for expressive high-quality open-weight TTS on large multilingual data, if you can accept less active maintenance and no editing.
Zonos-v0.1 is a leading open-weight text-to-speech model trained on more than 200k hours of varied multilingual speech, delivering expressiveness and quality on par with—or even surpassing—top TTS providers.
choose when inference speed (150x realtime) matters more than editing and multilingual coverage.
A high-quality rapid TTS voice cloning model that reaches speeds of 150x realtime.
choose if you want a Whisper-inverted TTS and are okay with an academic, less supported model without editing.
An Open Source text-to-speech system built by inverting Whisper.
choose when you want a broader speech/sound generation family, not just speech editing.
An open-source model family for long-form speech, dialogue synthesis, voice design, sound effects, and real-time streaming TTS
choose for highly controllable TTS with broad language coverage, if you don't need zero-shot cloning or editing.
Controllable and fast Text-to-Speech for over 7000 languages!
choose when you need emotion/style editing and robust zero-shot TTS from a 3B LLM, despite higher compute.
A powerful 3B-parameter, LLM-based Reinforcement Learning audio edit model excels at editing emotion, speaking style, and paralinguistics, and features robust zero-shot text-to-speech
choose for cross-lingual/zero-shot TTS tuned by Youdao, if speech editing is not required.
Confucius4-TTS: a Multilingual and Cross-Lingual Zero-Shot TTS Engine
choose for lightweight instant voice cloning with style control, if you don't need speech editing.
Instant voice cloning by MIT and MyShell. Audio foundation model.
choose for an industrial, controllable and efficient zero-shot TTS without speech editing.
An Industrial-Level Controllable and Efficient Zero-Shot Text-To-Speech System
choose for a full-stack multilingual TTS with training and deployment support, no speech editing.
Multi-lingual large voice generation model, providing inference, training and deployment full-stack ability.
choose for streaming TTS and free-form voice design, if editing is not needed.
Qwen3-TTS is an open-source series of TTS models developed by the Qwen team at Alibaba Cloud, supporting stable, expressive, and streaming speech generation, free-form voice design, and vivid voice cloning.
choose for a lightweight zero-shot TTS inference implementation, if you accept sparse updates and no editing.
Spark-TTS Inference Code
choose for zero-shot voice cloning in 600+ languages, if you don't need speech editing.
High-Quality Voice Cloning TTS for 600+ Languages
choose for an LLM-empowered foundation TTS, if editing is not a priority.
An Open-Sourced LLM-empowered Foundation TTS System
choose for controllable emotion-expressive zero-shot TTS via multi-reward RL, no speech editing.
GLM-TTS: Controllable & Emotion-Expressive Zero-shot TTS with Multi-Reward Reinforcement Learning
choose for human-sounding expressive TTS, if you can forgo speech editing.
Towards Human-Sounding Speech
choose for human-like expressive TTS from a foundational model, no speech editing.
Foundational model for human-like, expressive TTS
An earlier or more limited way to do the job. Still the right call when you need something small, proven, or CPU-only.
choose if you need real-time on-device Vietnamese TTS with cloning, accepting single-language scope.
Vietnamese TTS with instant voice cloning • On-device • Real-time CPU inference • 48kHz audio quality • Chuyển văn bản thành giọng nói tiếng Việt • Text to speech tiếng Việt • TTS tiếng Việt
choose for fast local TTS with pre-trained voices, not zero-shot, if you need offline CPU inference.
A fast, local neural text to speech system
These projects were analysed and named VoiceCraft among their alternatives. The relationship is not symmetric — how VoiceCraft rates them is a separate judgement, made when VoiceCraft is analysed in its own right.
calls VoiceCraft “choose when you also need zero-shot speech editing in addition to TTS.”
Instant voice cloning by MIT and MyShell. Audio foundation model.
calls VoiceCraft “choose for zero-shot TTS and speech editing in the wild.”
An Industrial-Level Controllable and Efficient Zero-Shot Text-To-Speech System
calls VoiceCraft “Choose when you need zero-shot TTS plus speech editing in the wild, with an LLM-based approach similar to CosyVoice.”
Multi-lingual large voice generation model, providing inference, training and deployment full-stack ability.
calls VoiceCraft “Choose when you also need zero-shot speech editing in addition to TTS.”
Qwen3-TTS is an open-source series of TTS models developed by the Qwen team at Alibaba Cloud, supporting stable, expressive, and streaming speech generation, free-form voice design, and vivid voice cloning.
calls VoiceCraft “choose over Zonos if you also need zero-shot speech editing in addition to TTS; both are token-based generation models, with VoiceCraft emphasizing in-the-wild editing.”
Zonos-v0.1 is a leading open-weight text-to-speech model trained on more than 200k hours of varied multilingual speech, delivering expressiveness and quality on par with—or even surpassing—top TTS providers.
calls VoiceCraft “choose when you also need zero-shot speech editing, not just TTS.”
Spark-TTS Inference Code
calls VoiceCraft “Use for zero-shot TTS plus speech editing in one model, rather than pure TTS.”
High-Quality Voice Cloning TTS for 600+ Languages
calls VoiceCraft “Choose for token-infilling zero-shot speech editing and TTS when you need content edits rather than emotion/style RL edits.”
A powerful 3B-parameter, LLM-based Reinforcement Learning audio edit model excels at editing emotion, speaking style, and paralinguistics, and features robust zero-shot text-to-speech