whisperX Releases
103 releases of m-bain/whisperX
- v3.7.7
A performance-focused update that optimizes speaker assignment and fixes timestamp formats for subtitle exports.
Feb 13, 2026
- v3.7.6
The project removes support for Python 3.9 to align with modern dependency management practices.
Jan 27, 2026
- v3.7.5
This release introduces CLI hotwords for better recognition of rare terms and improves documentation for troubleshooting.
Jan 27, 2026
- v3.7.4
Core deep learning dependencies are upgraded to the latest versions to ensure performance and stability.
Oct 16, 2025
- v3.7.3
Adds support for Swedish language alignment and stabilizes PyTorch version constraints to prevent conflicts.
Oct 16, 2025
- v3.7.2
The installation process is refined to restrict the Triton dependency to specific x86_64 Linux architectures.
Oct 12, 2025
- v3.7.1
Dependency constraints are updated to ensure full compatibility with Python 3.13.
Oct 12, 2025
- v3.7.0
The primary feature of this version is enabling the project to run on the latest Python version.
Oct 10, 2025
- v3.6.0
This release focuses on code organization and maintenance, including a file rename to avoid standard library conflicts and the introduction of centralized logging to replace ad-hoc print statements.
Oct 10, 2025
- v3.5.0
This update improves text normalization with expanded abbreviation support and updates dependencies to include the latest PyTorch and CUDA versions.
Oct 8, 2025
- v3.4.3
A maintenance release addressing critical bugs in the Voice Activity Detection (VAD) system and resolving dependency compatibility issues.
Oct 1, 2025
- v3.4.2
A bug fix release that resolves an IndexError occurring during emission calculation by ensuring proper integer tensor indexing.
Jun 27, 2025
- v3.4.1
A quick hotfix designed to correct functionality issues specifically related to speaker embedding.
Jun 25, 2025
- v3.4.0
This release enhances diarization capabilities and improves project documentation and workflow stability.
Jun 24, 2025
- v3.3.4
A performance-focused update that optimizes the startup time for the Command Line Interface (CLI).
May 3, 2025
- v3.3.3
A feature-rich release introducing Silero VAD support, new language models, and a migration to the `uv` package manager.
May 1, 2025
- v3.3.2
A patch release aimed at resolving CUDNN related issues by updating the underlying CTranslate2 version requirement.
Apr 10, 2025
- v3.3.1
A minor release focusing on developer experience improvements through type hints and better text output during diarization.
Jan 8, 2025
- v3.3.0
This release focuses on dependency updates (faster-whisper), enabling the distil-large-v3 model, and adding offline capabilities via the local_files_only option. It also includes infrastructure improvements like build workflows and installation testing.
Jan 2, 2025
- v3.2.0
This release significantly expands language support by adding alignment and ASR models for several languages including Korean, Czech, and Romanian. It also includes various bug fixes related to VAD, diarization, and transcription stability, alongside documentation updates and performance optimizations.
Dec 18, 2024