FunASR Releases

Sep 9, 2026

This release focuses on NumPy 2 compatibility, enhances streaming and training reliability, and expands MOSS deployment capabilities with new examples and documentation.

Sep 9, 2026

This release focuses on fixing portable source archives for Git archive and source installs, alongside shipping updated deployment guidance for real-time and industrial scenarios. It includes verified cross-platform binaries for the runtime-llamacpp component alongside the standard wheel and source packages.

Sep 3, 2026

This release introduces support for n8n OpenAI audio transcription, fixes issues related to ONNX runtime dependencies and NumPy 2 ABI mismatches, and improves real-time VAD handling. It also includes documentation updates and new runtime downloads for various platforms.

Sep 2, 2026

FunASR v1.4.12 focuses on stabilizing runtime performance and expanding deployment options. Key updates include fixes for real-time partials and vLLM decoding, alongside the addition of new runtime binaries for various platforms.

Sep 1, 2026

FunASR v1.4.11 is a maintenance release that fixes SentencePiece word markers alignment in the auto-model and provides pre-built runtime binaries for various platforms. The update includes self-contained llama.cpp binaries for Windows, Linux, and macOS, available for installation via PyPI.

Aug 30, 2026

This release improves subtitle segmentation for long Chinese sentences by balancing duration and length while preferring lexical boundaries and punctuation. It also includes verified llama.cpp runtime archives for multiple platforms and hardware backends.

Aug 30, 2026

This release focuses on bug fixes for subtitle generation, auto-model timestamp handling, Paraformer padding, and realtime segment preservation. It also includes documentation updates for MOSS diarization and provides new prebuilt runtime binaries for various platforms, including support for Windows CUDA Blackwell.

Aug 30, 2026

This release expands Moss model capabilities by adding support for vLLM diarized responses and long transcriptions. It also introduces Windows CUDA assets targeting the Blackwell architecture (sm_120) and fixes a download snapshot error.

Aug 30, 2026

This runtime release provides prebuilt binaries for the llama.cpp backend, specifically adding support for the Blackwell GPU architecture and Vulkan backends on Windows and Linux.

Aug 30, 2026

This maintenance release addresses critical runtime decoding issues for SenseVoice and introduces performance profiling tools for real-time processing.

Aug 29, 2026

This runtime release fixes a critical Vulkan weight upload bug and provides updated prebuilt binaries for various platforms including Vulkan and CUDA.

Aug 29, 2026

This release provides prebuilt self-contained binaries for the FunASR llama.cpp / GGUF runtime, supporting SenseVoice, Paraformer, and Fun-ASR-Nano with built-in FSMN-VAD, offering cross-platform options including Vulkan and CUDA backends.

Aug 29, 2026

This Python package release bundles the llama.cpp runtime v0.2.3 and includes documentation updates, bug fixes for timestamp generation and real-time timeouts, and a new CLI feature for grouping subtitle cues.

Aug 29, 2026

This release enhances diagnostic capabilities for the llama.cpp runtime by extending SenseVoice error reporting with granular stage boundaries, allowing users to pinpoint failures during backend initialization, model loading, or graph execution. It also prepares Windows Vulkan packages for specific AMD GPU retesting.

Aug 28, 2026

This release focuses on improving runtime stability and error visibility by guarding backend initialization failures and flushing stderr boundaries to clearly distinguish between backend selection, initialization, and buffer resolution stages.

Aug 28, 2026

This release simplifies the default Python inference installation by removing `torchaudio` as a hard dependency and introducing a `kaldi-native-fbank` backend. It also bundles prebuilt llama.cpp runtimes alongside the Python distributions to streamline deployment.

Aug 27, 2026

Introduces prebuilt self-contained binaries for the FunASR llama.cpp runtime, supporting SenseVoice, Paraformer, and Fun-ASR-Nano models with built-in FSMN-VAD. It offers various hardware optimizations including AVX2, Vulkan, and CUDA backends to enhance performance across different platforms.

Aug 26, 2026

This release focuses on major performance improvements for Realtime WebSocket concurrency and enhanced compatibility. It removes serialization locks, batches decode requests, and improves exception handling while maintaining compatibility with older PyTorch versions.

Aug 26, 2026

FunASR 1.4.3 introduces an optional Silero VAD adapter with configurable thresholds and ONNX support, alongside optimizations for speaker embedding clustering. The release refreshes ecosystem integrations and includes verified prebuilt llama.cpp runtimes for Linux, macOS, and Windows.

Aug 21, 2026