SeaLLMs-Audio-7B

SeaLLMs-Audio-7B is the multimodal audio extension of the SeaLLMs family and, by its authors' account, the first large audio-language model tailored for multiple Southeast Asian languages. It supports Indonesian, Thai, and Vietnamese alongside English and Chinese, accepting audio-only, text-only, or combined audio-text input.

The model covers a wide task range: audio captioning, speech recognition, speech-to-text translation, emotion recognition, spoken question answering, speech summarization, and voice-based dialogue over factual and general-knowledge queries. The team also released SeaBench-Audio, a benchmark for evaluating audio-language models on Southeast Asian languages. License terms should be verified on the model card before commercial use.