2 Releases published by 1 person

• 1.1.6

  published Jul 2, 2025

• 1.1.7

  published Jul 14, 2025

1 Pull request merged by 1 person

• Update WAV File Naming and Dependencies 📝🔊

  #1091 merged Jun 24, 2025

2 Pull requests opened by 2 people

• feat: Add subtitle generation with speaker and timing information

  #1124 opened Jul 12, 2025

• path in database: absolute -> relative

  #1134 opened Jul 16, 2025

30 Issues closed by 12 people

• The generated audio cannot be played, there is a warning, the interface cannot be clicked to play, and there is no sound wave

  #1137 closed Jul 17, 2025

• Unable to read the data I prepared

  #1131 closed Jul 17, 2025

• Feature request: train benchmark

  #1128 closed Jul 14, 2025

• Why is my (Gradio App) webpage blank? help me

  #1129 closed Jul 14, 2025

• 生成的语音跟念经一样

  #1127 closed Jul 13, 2025

• Empty page in browser using `f5-tts_infer-gradio`

  #1126 closed Jul 12, 2025

• The train code not compatible with latest accelerate 1.8.1, please fix it to 1.6.0

  #1111 closed Jul 8, 2025

• Can anyone explain why the attn_mask_enabled option is set to False?

  #1118 closed Jul 8, 2025

• F5-TTS repo and pinokio versions fail to synthesize cloned voice/ regular user-not a programmer

  #1117 closed Jul 7, 2025

• help pls

  #1092 closed Jul 4, 2025

• Numbers normalisation

  #1065 closed Jul 4, 2025

• [Doc] Official code from FrontierLabs for the paper "F5R-TTS: Improving Flow Matching based Text-to-Speech with Group Relative Policy Optimization"

  #957 closed Jul 4, 2025

• Can someone merge A-DMA to accelerate finetuning?

  #1116 closed Jul 4, 2025

• generated wav has part of ref wav

  #1114 closed Jul 4, 2025

• gradio root_path configuration not effective in f5-tts_infer-gradio

  #1093 closed Jul 4, 2025

• I have no idea if its a bug bull ill report it anyways.

  #1085 closed Jul 4, 2025

• Glitching in the synthesized output

  #1079 closed Jul 4, 2025

• Paying for a developer

  #1080 closed Jul 4, 2025

• I found some bug

  #1075 closed Jul 4, 2025

• Help setting speed using CLI with multispeech

  #1112 closed Jul 3, 2025

• 请求帮助，不会英文，所以中文发言，感谢。

  #1110 closed Jul 1, 2025

• MMDiT streams difference

  #1107 closed Jun 28, 2025

• 需要帮助，推理短文字，会跳过。

  #1108 closed Jun 28, 2025

• voice-chat bug

  #1082 closed Jun 28, 2025

• 报下面的错误，也可以正常推理，麻烦帮忙看下。

  #1106 closed Jun 26, 2025

• 推理中文出来，如果遇到数字会说英文，不会推理阿拉伯数字。

  #1103 closed Jun 25, 2025

• How to use custom models

  #1102 closed Jun 24, 2025

• 请问这是什么错误

  #1099 closed Jun 24, 2025

• Is this current version compatible with the 5090 graphics card?

  #1098 closed Jun 23, 2025

• triton-inference-server搭建加速环境

  #1089 closed Jun 21, 2025

14 Issues opened by 12 people

• F5-TTS repo and pinokio versions fail to synthesize cloned voice/ regular user-not a programmer

  #1136 opened Jul 16, 2025

• Can this project transfered into tensorrt fp16 format safely?

  #1135 opened Jul 16, 2025

• Change absolute path to relative in `raw.arrow`

  #1133 opened Jul 16, 2025

• Batch process: one GPU, but need to generate multiple voice files at once

  #1123 opened Jul 12, 2025

• Single-Speaker Training Produces Unintelligible Text Output

  #1122 opened Jul 11, 2025

• Facing issue during cross-sentence evaluation [SIM] of LibriSpeech-PC test-clean dataset.

  #1121 opened Jul 9, 2025

• api.py generates normal audio with x86, but noise output with aarch64; infer-cli both fine

  #1115 opened Jul 2, 2025

• 知识付费，请求搭建F5-TTS-triton_trtllm

  #1113 opened Jul 1, 2025

• Artifacts and Noise in Fine-Tuned Model

  #1105 opened Jun 26, 2025

• Blank Audio getting generated in case of using TRT optimization

  #1104 opened Jun 25, 2025

• Long Silence in Generated Speech After ~800k Steps

  #1101 opened Jun 24, 2025

• Query regarding the amount of data used for training F5-TTS Small

  #1097 opened Jun 20, 2025

• Conditioning on 54 emotion scores and freestyle emotion captions

  #1096 opened Jun 20, 2025

• Couldn't assign noise with shape torch.Size([3660, 100]), engine supports [min, opt, max] = [(2, 100, 100), (2, 1500, 100), (8, 3000, 100)]

  #1094 opened Jun 20, 2025

5 Unresolved conversations

Sometimes conversations happen on old items that aren’t yet closed. Here is a list of all the Issues and Pull Requests with unresolved conversations.

• Supported Languages?

  #87 commented on Jun 19, 2025 • 0 new comments

• Using TRT-LLM for acceleration, with high concurrency, severe problem observed

  #1039 commented on Jun 30, 2025 • 0 new comments

• The beginning of the synthesized audio contains other speech

  #1041 commented on Jul 9, 2025 • 0 new comments

• Explicit pauses

  #1069 commented on Jul 14, 2025 • 0 new comments

• Question about TensorRT-LLM inference acceleration

  #967 commented on Jul 17, 2025 • 0 new comments