How to use F5 TTS: what you can do with it

3 use cases

• Generate speech in a reference voice from the CLI

  Install the package and run one command with your reference audio and the text you want spoken.

  • REPO SWivid/F5-TTS
  • Claude Code
  • Codex
  • Cursor
  • Gemini CLI
  • Content
  • Intermediate
  View prompt

  Using SWivid/F5-TTS, set up the environment, install `pip install f5-tts` and run f5-tts_infer-cli with my reference clip and transcript to read <my script>, saving the audio.

  Share Email me this

• Run a local web app for multi style and multi speaker audio

  Launch the Gradio interface to generate chunked, multi speaker or voice chat audio in a browser.

  • REPO SWivid/F5-TTS
  • Claude Code
  • Codex
  • Cursor
  • Gemini CLI
  • Content
  • Beginner
  View prompt

  With SWivid/F5-TTS, start the Gradio app on port 7860 and walk me through producing a two speaker dialogue from <my script>.

  Share Email me this

• Run F5 TTS in a container instead of installing it

  Use the published image or build the Dockerfile so the CUDA and PyTorch setup is fixed.

  • REPO SWivid/F5-TTS
  • Claude Code
  • Codex
  • Cursor
  • Gemini CLI
  • Content
  • Intermediate
  View prompt

  Using SWivid/F5-TTS, get the docker image running with the hub cache volume and port 7860, then generate one sample to prove the setup works.

  Share Email me this

What people are building with F5 TTS

• Built withthe web

  Narrated a whole novel locally with F5-TTS and a casting system

  He built a multi-voice audiobook pipeline on an 8GB consumer GPU, calling the F5TTS API with a 10 second reference clip and its transcript for each line, and using only public domain seeds or his own voice. The lesson he reports is that consistency beats per-voice polish: a pinned seed, fixed speed, one loudness target and short crossfades stop the read from lurching between clips.

  F5 TTS on GitHub

  By CryptoJones (cryptojones.dev) on the web·Jun 7, 2026

  View original post on the web

• Hands-onthe web

  Cloned a voice with F5-TTS from 12 seconds and wrote the post-mortem

  He ran f5-tts_infer-cli with F5TTS_v1_Base on an RTX 2080 Ti using a 12 second clip pulled from a Telegram voice message. The output was recognizable but flat, and he traces it to the reference: codec artifacts and background noise got copied, and he now recommends two to three minutes of clean uncompressed audio before trying again.

  F5 TTS on GitHub

  By Amine El Farssi (amineelfarssi.github.io) on the web·Mar 1, 2026

  View original post on the web

Real builds shared on X, Hacker News, Reddit, DEV and GitHub, summed up by Deduped from each post and its links; each card links to the original post. See what people are building with other repos

Quick start with F5 TTS

1. Check you have what it needs: Python 3.10 or newer, for example a conda env created with `conda create -n f5-tts python=3.11`; PyTorch and torchaudio matched to your device, plus FFmpeg; A reference WAV clip of the voice you want, and its transcript if you do not want ASR to transcribe it.
2. Open SWivid/F5-TTS on GitHub and read the README and license.
3. Create an environment with conda create -n f5-tts python=3.11 and conda activate f5-tts, then install FFmpeg with conda install ffmpeg
4. Install PyTorch and torchaudio for your device as the README lists, then install the package with pip install f5-tts
5. Run f5-tts_infer-cli --model F5TTS_v1_Base with --ref_audio, --ref_text and --gen_text set to your clip, its transcript and your text
6. Leave --ref_text "" to have an ASR model transcribe the clip instead, at the cost of extra GPU memory

These steps come from our top use case, Generate speech in a reference voice from the CLI, which also has a copy-paste agent prompt.

Using the Deduped MCP, get the setup steps for SWivid/F5-TTS

claude mcp add --transport http deduped https://dedupedai.com/mcp

Best AI repos for Content

About this repo

• Who it is forDevelopers and creators who want open source voice generation they can run themselves.
• Why it is hot15k stars, picked up by one AI source we track
• Sourcegithub.com/SWivid/F5-TTS · Back to Hot repos

Requirements

• Python 3.10 or newer, for example a conda env created with `conda create -n f5-tts python=3.11`
• PyTorch and torchaudio matched to your device, plus FFmpeg
• A reference WAV clip of the voice you want, and its transcript if you do not want ASR to transcribe it
• A GPU for comfortable speed; the published benchmarks use a single L20

Related repos

• Trended earlier REPO KittenML/KittenTTS Kitten TTS Kitten-TTS is an open-source family of text-to-speech models, including the 1.7B 1-bit Kitten-TTS 2 with voice cloning and expression control and lightweight 15M to 80M models that run on CPU.
• Trended earlier REPO Sentinel3252/Native-Breeze-TTS-2-Support-for-vLLM-Omni Native Breeze TTS 2 Support for vLLM Omni A learning project that adds Breeze TTS 2 speech synthesis support to the vLLM-Omni serving framework, with the official reference code and the modified framework source side by side.
• Trended earlier REPO hexgrad/kokoro Kokoro kokoro is the inference library for Kokoro-82M, an open-weight text-to-speech model with 82 million parameters that rivals larger models while being faster and cheaper to run. A KPipeline turns text into 24 kHz audio with voices like af_heart, and supports American and British English, Spanish, French, Hindi, Italian, Japanese, Brazilian Portuguese and Mandarin.