Question: commercial use of SenseVoiceSmall weights under MODEL_LICENSE (Android app, incl. users in China)

Aberta
  • Algo errado?
#3,795 0 comentários 0 reações 0 responsáveis Ver no GitHub

Mantenedores costumam responder em até 1 dia

Ninguém assumiu esta issue ainda.

Avaliação

Dificuldade
1/5
Tempo estimado
Menos de uma hora
Facilidade para iniciantes
20/100
Tipo de issue
Documentação
Clareza
Claramente especificada
Status de atividade
Ativa
Stack de tecnologia
python
Domínio
documentation

Direção de pesquisa

The question concerns SenseVoiceSmall's MODEL_LICENSE and its Hugging Face model card. Start by reviewing the license text and the linked Hugging Face discussion; the issue is resolved when the team clarifies commercial use, redistribution and attribution, and any relevant restrictions.

Escrita pelo modelo de indexação a partir do texto da issue.

Descrição

needs triage question

Hi FunASR team,

I'm an independent developer building an offline travel interpreter app for Android (Korean / Chinese / Japanese / English). I would like to use SenseVoiceSmall (ONNX int8, run via sherpa-onnx) for fully on-device speech recognition.

The Hugging Face model card lists "license: other / model-license", which points to the MODEL_LICENSE file in this repository. I posted the same question there (https://huggingface.co/FunAudioLLM/SenseVoiceSmall/discussions/18) but have not received an answer yet.

Could you please clarify:

  1. Commercial use: May the SenseVoiceSmall weights be used inside a commercial app (free or paid) distributed through app stores, including to users in mainland China?
  2. Redistribution: May I bundle the converted (ONNX/int8) weights in the app package, or download them on first run? If so, what attribution (model name, source, license text) is required?
  3. Wording of the license: The license text mentions "reference / learning purposes" (section 3). Does this restrict use of the model inside an end-user application?
  4. Are there any restrictions from the training data or other upstream components that I should be aware of?

I only plan to ship the pretrained weights, not any training data. A short written answer here would help me decide whether I can ship it. If this is the wrong place to ask, please point me to the right channel.

Thank you!

Linguagem predominante
Python
Estrelas
21k
Forks
2.1k
Merge médio
15h 16min
PRs com merge (30d)
85

Preparar o ambiente

Primeiros passos

  1. Leia a issue inteira e depois o guia de contribuição do projeto.
  2. Comente na issue dizendo que vai assumir — evita que duas pessoas façam o mesmo trabalho.
  3. Faça um fork do repositório e trabalhe em uma branch.
  4. Abra um pull request que referencie o número da issue.

Mais de modelscope/FunASR

Todas as issues de modelscope/FunASR

Issues semelhantes

Mais issues de Python

Receba novas issues na sua caixa de entrada

Um resumo curto de issues do GitHub para quem está começando.