MOSS-Audio Releases

A major feature update that introduces a comprehensive onboarding experience, expands model capabilities with redundancy options, and improves workflow interactivity through loop edges and client-side polling.

Apr 19, 2026

A maintenance release focused on Safari browser compatibility, fixing rendering issues with text, shadows, and opacity, along with build and CDN updates.

Apr 10, 2026

A minor maintenance update that corrects the execution order of specific triggers and updates the build and CDN links.

Apr 10, 2026

This release enables audio-driven video generation workflows and introduces a dynamic array batch mode for sequential processing. It also significantly optimizes memory usage during undo/redo operations.

Apr 1, 2026

This release introduces a comprehensive Video Input node and a full Undo/Redo history system to streamline workflow creation. It also enhances the user interface with Veo model parameters and integrates a waitlist feature for the NB Pro tier.

Mar 30, 2026

Users can now manage saved workflows with the new browser and improved media externalization, keeping assets portable. The update also adds optional input logic and a redesigned group context menu, alongside performance improvements.

Mar 29, 2026

This is a maintenance release focused on stability, specifically addressing a potential crash that could occur during drag-and-drop interactions involving transformers.

Mar 16, 2026

A minor maintenance release designed to correct structural issues that could result in incorrect behavior within the application.

Mar 13, 2026

Feb 7, 2026

This release adds CLI revision options, introduces a new safe transform method for calibration, fixes Lightning 2.6+ compatibility, and updates the pyannote-database dependency.

Dec 7, 2025

This update introduces hardware support for Mac M* and CPU MPS, switches model formats from PyTorch to Safetensors, and adds advanced sampling features like Conditional Free Guidance (CFG) and Min_P Sampler.

Jun 13, 2025

Introduces a new graphical user interface for precise control over pitch and duration, allowing users to fine-tune utterances and switch voices across 7000+ languages.

Oct 7, 2024

Significantly improves audio quality by retraining the vocoder end-to-end on synthetic spectrograms and increasing its capacity, without slowing down inference speed.

Sep 22, 2024

Enhances the TTS system by introducing stochastic prosody prediction and expanding language support, resulting in more natural-sounding speech.

Jul 25, 2024

A major milestone release enabling text-to-speech synthesis in over 7000 languages through model extrapolation, alongside improved quality and a new watermarking feature.

Jun 10, 2024

Adds emotional control to text-to-speech by allowing models to be conditioned on emotional prompts during training and transferring that emotion to generated speech.

Jun 10, 2024

Apr 15, 2019

Jan 16, 2019