ElevenLabs alternatives for TTS and voice cloning

Chatterbox TTS Server is the nearest match for everyday text-to-speech: built-in voices, voice cloning from a WAV or MP3 clip, and long texts split and joined into one file for narration. It runs on your own computer or server with a browser interface.

It speaks English with emotion control and other languages through Chatterbox Multilingual, and offers an OpenAI-compatible API for apps. It runs on NVIDIA, AMD, Apple Silicon or a CPU.

Voice-Clone-Studio combines reference-voice generation with audio preparation and scripts for multiple speakers. Qwen3-TTS can also design a voice from a written description, while several speech engines share the same workspace.

GPT-SoVITS is for voice cloning. It generates speech from a short reference recording or fine-tunes a model on a custom voice, in English, Japanese, Korean, Cantonese and Chinese, including a language other than the recording's.

The browser interface includes tools to clean and split training audio.

TTS WebUI gathers many local speech models in one browser interface: Bark, Tortoise and StyleTTS2 by default, with Kokoro, XTTSv2, CosyVoice and others as extensions. It also covers voice conversion with RVC, transcription with Whisper, and music generation.

Its OpenAI-compatible speech API connects it to Open WebUI and SillyTavern. Model weights have their own licenses.

Speech AI Forge puts several speech engines in one browser interface for narration and voice experiments. You can choose speakers or supply reference audio, then use the separate API server when another app needs the same speech generation.

IndexTTS offers voice cloning with separate controls for delivery and speaker identity. Its browser interface can use one recording for the voice and another for emotion, with pronunciation and speaking-speed adjustments.

LocalText2Voice suits audiobooks and long scripts that need chapter structure and editable narration segments. You can regenerate one passage, change voices and export timed subtitles or mix narration with background music.

Alexandria Audiobook suits narrated books with separate character voices. Its editor lets you correct speaker assignments, preview lines and regenerate passages before exporting a chaptered audiobook. Script annotation needs a separate language model.

Voice-Pro combines recording preparation, transcription and voice cloning in a browser workspace. Generate speech from reference audio or prepare translated dubbing with its supported engines; the application requires internet access.

F5-TTS covers speech generation through a browser interface, with longer passages split into chunks and several speakers or speaking styles in one output. Its supplied pretrained models are restricted to noncommercial use.

pyVideoTrans replaces AI dubbing. It transcribes a video, translates it and produces a dubbed version with subtitles, pausing for you to proofread at each stage. Speaker separation gives each person in a conversation their own voice, and voice cloning works with F5-TTS, CosyVoice or GPT-SoVITS.

With local models such as Faster-Whisper and Ollama, the whole pipeline can run offline.

TTS-Audio-Suite builds speech workflows in ComfyUI for character dialogue and subtitle-timed narration. Generate speech against SRT timings, switch voices within a script and regenerate only the segments you change, using the selected speech engines.

Applio handles voice changing. It converts recordings or live microphone input into another voice, using community models or ones you train on your own audio, and includes text-to-speech you can pass through a voice model.

It needs no account and collects no telemetry.

Kokoro-FastAPI covers the speech API part of ElevenLabs for developers building narration or reading apps. It streams speech, mixes voices and returns caption timestamps, with controls for pauses and pronunciation.

ChatTTS Colab suits scripts with different character voices. You can audition batches of voice candidates, save them for reuse and generate long passages with streaming playback.

CosyVoice suits developers replacing a speech-generation service with their own deployment. It combines reference-voice synthesis, multilingual output and streaming audio, with API interfaces for applications.

XTTS v2 suits developers who need a speech server that clones a reference voice and can speak another supported language. Coqui TTS provides command-line and server access, and saved cloned voices can be reused without supplying the sample each time.

ebook2audiobook covers audiobook narration. It turns non-DRM EPUB, MOBI, PDF and Word files into audiobooks with chapters and metadata, and can narrate in a voice cloned from your own recording.

It runs locally with a browser interface and outputs M4B, MP3 or other formats. Languages depend on the speech engine you pick.

Abogen turns books and scripts into narrated audio with chapter metadata and matching captions. Timestamped input can guide narration timing, but this is a narration workflow rather than reference-voice cloning.

SmartSub covers video dubbing with generated speech matched to subtitle timing. Edit or translate the subtitles, choose speech synthesis or voice cloning, then export the video with its dubbing track.

audiblez is a simple way to turn EPUB books into M4B audiobooks with Kokoro voices on your own computer. Pick a narrator, reading speed and chapters in the graphical app or on the command line.

Voices cover English, Spanish, French, Hindi, Italian, Japanese, Brazilian Portuguese and Mandarin Chinese.