GitHub / FunAudioLLM/CosyVoice issues and pull requests
#1877 - 首包出现重复token或幻觉问题
Issue -
State: closed - Opened by dl8207531 6 months ago
- 4 comments
Labels: stale
#1840 - Congratulations on the Fun-CosyVoice 3.5 Release!
Issue -
State: closed - Opened by linyueqian 8 months ago
- 9 comments
Labels: stale
#1598 - [Runtime] StepAudio2 Streaming DiT Token2Wav Integration
Pull Request -
State: open - Opened by yuekaizhang about 1 year ago
#1596 - Issue with number reading in triton_trtllm deployment
Issue -
State: open - Opened by Xiaomuma001 about 1 year ago
#1594 - 流式生成质量问题
Issue -
State: open - Opened by cainiaobo about 1 year ago
#1593 - 某些英文字母的发音错误
Issue -
State: open - Opened by liberals1 about 1 year ago
#1592 - 請問cosyvoice 3有開源的時間表嗎. 十分期待
Issue -
State: open - Opened by RockyFong about 1 year ago
#1589 - 这是阿里开源的Qwen3-TTS吗
Issue -
State: open - Opened by Hommy-master about 1 year ago
#1588 - 安装镜像报错: Could not open requirements file: [Errno 2] No such file or directory: 'requirements.txt'
Issue -
State: open - Opened by wangyf6673 about 1 year ago
#1585 - ttsfrd如何添加韩语支持
Issue -
State: closed - Opened by xhx1304 about 1 year ago
#1580 - pip install -r requirements.txt 报错
Issue -
State: closed - Opened by ganluo960214 about 1 year ago
#1579 - cosyvoice使用已注册id流式生成克隆语音,前两秒是空白的。
Issue -
State: open - Opened by yesulove about 1 year ago
#1578 - 克隆音色应该怎样训练
Issue -
State: closed - Opened by xhx1304 about 1 year ago
- 2 comments
#1575 - 目前看在调用token2wav方法进行推理的时候,flow.inference只支持单batch推理,有考虑过改造成多batch提高QPS吗?
Issue -
State: open - Opened by makabaka6338 about 1 year ago
#1573 - 安装完之后卡在运行webui这一步,只好自己写了一个脚本来处理
Issue -
State: open - Opened by zero456 about 1 year ago
#1571 - 根据官方的环境搭建流程,总是报no module named 'conformer'
Issue -
State: open - Opened by Robert2045 about 1 year ago
#1570 - Using the latest cosyvocie 2-0.5B fine-tuning to train the big model, failed to train the hifigan model
Issue -
State: open - Opened by exceedzhang about 1 year ago
#1569 - 关于使用LibriTTS微调CV2的性能问题
Issue -
State: open - Opened by hariiiseldon about 1 year ago
#1564 - [Bug] Unwanted Background Music Appears in Korean Speech Synthesis
Issue -
State: open - Opened by jyjnote about 1 year ago
#1562 - Porting to lama.cpp
Issue -
State: open - Opened by EmreOzkose about 1 year ago
#1560 - 微调flow报错,但是微调llm没问题
Issue -
State: closed - Opened by 11075225 about 1 year ago
#1559 - webui生成音频时提示JSON相关报错
Issue -
State: open - Opened by staryzhou about 1 year ago
#1558 - 方言好像不生效啊
Issue -
State: open - Opened by hjj-lmx about 1 year ago
#1557 - 根据run.sh文件生成的parquet.tar文件打不开,显示文件损坏
Issue -
State: open - Opened by 11075225 about 1 year ago
#1555 - Bug: KeyError when accessing speaker embeddings in frontend_sft()
Issue -
State: open - Opened by ditingdapeng about 1 year ago
#1554 - 流式输入是否可以调用本地部署的cosyvoice 服务
Issue -
State: open - Opened by lion000 about 1 year ago
#1553 - gradio 与 kaldifst 冲突
Issue -
State: open - Opened by yao978318542 about 1 year ago
#1552 - No module named 'ttsfrd
Issue -
State: closed - Opened by yao978318542 about 1 year ago
#1551 - ValueError: /root/.cache/modelscope/hub/models/iic/CosyVoice2-0___5B/cosyvoice.yaml not found!
Issue -
State: open - Opened by LaiosOvO about 1 year ago
#1549 - 修改prompt_text参数后,生成的语音胡言乱语
Issue -
State: open - Opened by a1422232790 about 1 year ago
#1548 - Brazilian Portuguese language support
Issue -
State: open - Opened by marcelov2 about 1 year ago
#1547 - fastapi 生成的 wav 无法播放
Issue -
State: closed - Opened by vipwhr about 1 year ago
#1544 - vllm部署
Issue -
State: closed - Opened by laohua-wang about 1 year ago
- 1 comment
#1539 - vllm推理prompt_text为何不支持方言?
Issue -
State: open - Opened by hjc3613 about 1 year ago
#1538 - Unable to reproduce Expresso dataset results reported in CosyVoice3 paper
Issue -
State: open - Opened by yileitu about 1 year ago
#1537 - Instruct 合成到底有哪些输入啊?
Issue -
State: closed - Opened by JohnHerry about 1 year ago
#1536 - CosyVoice-300M-Instruct生成语音耗时非常长
Issue -
State: closed - Opened by surpoloyang about 1 year ago
- 2 comments
Labels: stale
#1535 - train with dpo throw error: FileNotFoundError: [Errno 2] No such file or directory: 'data/dev-clean_reject/utt2speech_token.pt'
Issue -
State: open - Opened by honglyua-il about 1 year ago
#1533 - 重新训练flow模型,转换为trt后,合成结果全是噪声
Issue -
State: closed - Opened by WangGewu about 1 year ago
#1532 - 关于长文本合成的语音问题
Issue -
State: closed - Opened by luoyuyimeng about 1 year ago
- 7 comments
Labels: stale
#1531 - 文本标准化的包在哪
Issue -
State: open - Opened by laohua-wang about 1 year ago
#1530 - 关于为代码添加类型注解的提议
Issue -
State: open - Opened by yahiro-code about 1 year ago
#1528 - 请问flow模型 fp16推理溢出问题解决了吗?
Issue -
State: closed - Opened by JohnHerry about 1 year ago
- 4 comments
#1527 - 模型是不是无法做到openai兼容
Issue -
State: open - Opened by kelliaao about 1 year ago
#1526 - docker容器内运行异常
Issue -
State: closed - Opened by angel-heart about 1 year ago
- 2 comments
Labels: stale
#1523 - ‘-’ 读成 负
Issue -
State: open - Opened by lsh-boop about 1 year ago
#1522 - CosyVoice2-0.5B sucess, CosyVoice-300M error: qwen_pretrain_path
Issue -
State: open - Opened by xiaohoua about 1 year ago
#1521 - 关于 CV2 vllm 的流式速度问题
Issue -
State: open - Opened by shao-ssq about 1 year ago
#1519 - 请添加docker部署项目使用CosyVoice2-0.5B模型的方式,十分感谢
Issue -
State: open - Opened by angel-heart about 1 year ago
#1516 - Question about inference_cross_lingual
Issue -
State: open - Opened by CharlesNii about 1 year ago
#1515 - 跨语言复刻(原声为中文,目标为英文)稳定性极差
Issue -
State: closed - Opened by raozhongbo about 1 year ago
- 4 comments
Labels: stale
#1514 - SFT 训练的方式需要加prompt吗?
Issue -
State: open - Opened by sunnnnnnnny about 1 year ago
#1513 - DiffRO如何训练多任务打分器模型
Issue -
State: open - Opened by JohnHerry about 1 year ago
#1512 - 求一个50系显卡的依赖文件
Issue -
State: closed - Opened by wuwei1017 about 1 year ago
- 5 comments
#1511 - CosyVoice2训练问题:使用自定义Kaldi数据集在第一轮训练中输出噪音
Issue -
State: closed - Opened by littlestronomer about 1 year ago
- 6 comments
Labels: stale
#1510 - 同一份代码,同一段文字转语音,4090 2-3秒,A100、3090都要10来秒,独占显卡
Issue -
State: open - Opened by mifanvv about 1 year ago
#1509 - Clarification on the definition of “streaming mode” in CosyVoice
Issue -
State: open - Opened by jyjnote about 1 year ago
#1508 - 词表不对齐
Issue -
State: open - Opened by wyhsgitname about 1 year ago
#1507 - Dev/lyuxiang.lx
Pull Request -
State: closed - Opened by aluminumbox about 1 year ago
#1506 - 生成的声音结尾结束不自然
Issue -
State: open - Opened by Yanceye about 1 year ago
#1505 - Cosyvoice DPO训练笔记
Issue -
State: closed - Opened by ScottishFold007 about 1 year ago
- 6 comments
#1504 - sft use_spk-embedding flag
Issue -
State: closed - Opened by deyituo about 1 year ago
- 3 comments
Labels: stale
#1502 - Support for Vietnamese Language (Tiếng Việt)
Issue -
State: closed - Opened by cafe2404 about 1 year ago
- 2 comments
Labels: stale
#1501 - Why inference_instruct2 does not take the prompt_text as input?
Issue -
State: closed - Opened by jlmeunier about 1 year ago
- 2 comments
Labels: stale
#1500 - 支持方言“客家话”生成吗?
Issue -
State: closed - Opened by bianyiyan about 1 year ago
- 3 comments
Labels: stale
#1499 - 用官方代码保存的音色在使用时根本不对,自己用代码保存的与官方三秒极速复刻有什么区别吗
Issue -
State: closed - Opened by Thunder-shadow about 1 year ago
- 3 comments
Labels: stale
#1497 - 流式输出和文本instruction不兼容?
Issue -
State: open - Opened by jerermyyoung about 1 year ago
#1496 - 模型训练新语言后推理会产生 RuntimeError: sampling reaches max_trials 100 and still get eos when ignore_eos is True, check your input!
Issue -
State: closed - Opened by kwang1993 about 1 year ago
- 1 comment
#1494 - about custom pronunciation for proper nouns
Issue -
State: closed - Opened by jiephy about 1 year ago
- 2 comments
Labels: stale
#1493 - 为什么下载vllm需要的torch包显示没有2.7.0版本
Issue -
State: closed - Opened by wovhwhy about 1 year ago
- 5 comments
Labels: stale
#1492 - 离线部署失败
Issue -
State: open - Opened by 1105X about 1 year ago
#1488 - 部署
Issue -
State: closed - Opened by laohua-wang about 1 year ago
- 2 comments
Labels: stale
#1486 - 音色设计
Issue -
State: open - Opened by SinoMiles about 1 year ago
#1485 - vllm报错
Issue -
State: closed - Opened by Eddyhzz over 1 year ago
- 1 comment
#1484 - 模型性能评估
Issue -
State: open - Opened by ANHDY over 1 year ago
#1481 - 生成的音频是胡言乱语
Issue -
State: open - Opened by WangGewu over 1 year ago
- 22 comments
#1479 - vllm load 报错
Issue -
State: open - Opened by yangaking over 1 year ago
#1477 - Loosen restriction tensorrt
Pull Request -
State: closed - Opened by ShotaTanemura over 1 year ago
#1476 - python server.py的api调用生成出来的音频音色有问题
Issue -
State: open - Opened by jianhualll over 1 year ago
#1474 - Update frontend.py
Pull Request -
State: open - Opened by Clabiyau over 1 year ago
#1472 - cosyvoice2 inference_instruct2 extra instruct_text at beginning
Issue -
State: open - Opened by zooyf over 1 year ago
#1471 - 求助vllm启动报错找不到cosyvoice.vllm
Issue -
State: open - Opened by nichongqbi over 1 year ago
#1470 - Fine-tuning strategy for cross-lingual
Issue -
State: open - Opened by pokabookinflab over 1 year ago
#1468 - 如何调整cosyvoice2模型的文本分割长度?
Issue -
State: open - Opened by DuannHang over 1 year ago
#1467 - vllm所依赖的cuda和gcc的版本是多少?
Issue -
State: open - Opened by cenwurong001 over 1 year ago
#1466 - 自行添加新音色
Issue -
State: open - Opened by Thunder-shadow over 1 year ago
#1465 - Emilia fully annotated with EmoNet / Empathic Insight Voice emotion captions & scores - ready for training :)
Issue -
State: open - Opened by christophschuhmann over 1 year ago
#1464 - 环境部署整的我好懵
Issue -
State: closed - Opened by a1422232790 over 1 year ago
- 3 comments
#1463 - Cosyvoice2 GRPO RL training recipe
Issue -
State: open - Opened by yuekaizhang over 1 year ago
#1461 - CosyVoice2-0.5B在A100上利用vllm推理rtf0.6以上
Issue -
State: open - Opened by ygch over 1 year ago
#1459 - pip install disable cache
Pull Request -
State: closed - Opened by huiwq1990 over 1 year ago
#1458 - 流式输入有概率在生成音频里面直接复制原始prompt音频里面的部分段,非流式输入则不会出现
Issue -
State: open - Opened by wuqi930907 over 1 year ago
#1457 - Language tag usage in Korean fine-tuning data for improved cross-lingual synthesis
Issue -
State: closed - Opened by pokabookinflab over 1 year ago
- 2 comments
#1454 - 这个模型怎么部署啊
Issue -
State: closed - Opened by laohua-wang over 1 year ago
- 1 comment
#1450 - VLLM安装时,是否要调整requirements.txt里面的版本,不调整的话,会有包冲突的问题,导致安装不上,尤其是Numpy
Issue -
State: open - Opened by itunic over 1 year ago
#1449 - A serious bug in DPO implementation.
Issue -
State: open - Opened by fispresent over 1 year ago
#1447 - 【返回类型问题】构建部署后调用/inference_zero_shot接口,模型返回的类型是什么
Issue -
State: open - Opened by winteryx over 1 year ago
#1446 - 使用transformers==4.53.1版本,生成语音会混乱
Issue -
State: open - Opened by yxf0314 over 1 year ago
#1445 - CosyVoice flow模型的参数初始化
Issue -
State: open - Opened by JohnHerry over 1 year ago
#1444 - s2st pipeline
Pull Request -
State: open - Opened by Alexuan over 1 year ago