OpenMOSS-Team/MOSS-Audio-Tokenizer-Nano is a lightweight audio tokenizer designed for the MOSS TTS family. It converts speech and audio signals into discrete tokens suitable for language model-based generation. The model supports feature extraction pipelines in Transformers and is part of research on neural codecs for audio and speech applications.
In the Foundation models & chat space, MOSS Audio Tokenizer Nano takes a focused approach. Efficiently converting raw audio into discrete tokens for neural text-to-speech and audio generation models. It is built as an open-source project for developers. MOSS Audio Tokenizer Nano is open source under the Open Source license. MOSS Audio Tokenizer Nano is available on the web and API.
Behind MOSS Audio Tokenizer Nano is OpenMOSS-Team, and it first shipped in 2025. Key capabilities include audio tokenization, neural codec, and speech processing.
Summary written by a language model from the project’s public pages.
What PulseGate has recorded for this listing
Closest matches by what these projects do