sherlock Releases

This update introduces Arabic model details and a new F5TTS v1 Small training configuration. It also optimizes batch processing by removing an ineffective ThreadPoolExecutor and fixes issues in utils_infer.py.

Mar 24, 2026

Added 94 new templates and 47 CVEs, focusing on critical vulnerabilities such as RCE in Qwik and Langflow. The release also corrects PoC errors and refines detection logic to reduce false positives.

Mar 16, 2026

The app now remembers the selected audio device between sessions and improves CPU compatibility checks. It also fixes batch transcription issues with files lacking extensions and ensures background processes stop correctly.

Mar 13, 2026

This release focuses on project maintenance and reliability, adding a CONTRIBUTING.md guide and setting up CI/CD pipelines for tests, builds, and security.

Mar 13, 2026

Focuses on stability by updating the transcription engine and improving the clarity of failure messages to help users troubleshoot issues more effectively.

Mar 7, 2026

Introduces a new 'Stable Timestamps' mode designed to provide much more accurate subtitle timing for videos, tutorials, and movies.

Mar 6, 2026

This release fixes macOS system audio recording, improves structured error messages for clearer failures, and includes internal improvements and dependency updates.

Mar 4, 2026

This release introduces Flash Attention support for MMDiT, adds a `show_info` parameter for preprocessing, and includes a fused AdamW option with a warning about attention mask memory usage.

Mar 4, 2026

This major update introduces audio import and retranscription features, improves transcription quality with parallel resampling, introduces a beta feature gating system, and improves memory management.

Mar 3, 2026

This update introduces a configurable recording save path, improves macOS system audio recording reliability, adds a 'Wall of Love' section, remembers recent languages, and adds a copy button for commands.

Mar 1, 2026

This release features a major engine and API overhaul, adds OpenAI API compatibility, expands prebuilt model support, improves WebGPU performance, and integrates XGrammar.

Feb 18, 2026

This update focuses on fixing speech editing artifacts, adding Latvian model support, supporting `hf://` links in CLI, and optimizing training with mel domain padding and torch utils checkpointing.

Feb 16, 2026

This release fixes issues with Whisper models (metadata, URL, size), resolves Windows audio device selection bugs, fixes OpenAI API connection tests, and implements auto-fetching models from LLM providers.

Feb 11, 2026

This major update introduces custom OpenAI integration, streamlines the onboarding process, and adds over-the-air updates and transcript recovery to enhance reliability and user flexibility.

Dec 31, 2025

This is an automatically generated release for tag v4.2.0 with no specific changelog details provided in the release notes.

Dec 11, 2025

This major release transforms Meetily into a standalone application with a unified architecture, offering 3x faster transcription using the Parakeet model, pause/resume recording capabilities, and advanced AI summary templates.

Oct 23, 2025

Minimal release notes: "Automatically generated release for refs/tags/v4.0.2"

Sep 18, 2025

This release significantly reduces false positives through automatic testing and filtering of prone sites. It adds new CLI flags like --json for PR numbers and --no-txt, achieves PEP 561 compliance with mypy support, and deprecates Python 3.8 and 3.9.

Sep 16, 2025

This release introduces Mac MPS support, updates model formats to safetensors, and fixes CUDA errors. It also enhances sampling capabilities by adding the Min_P Sampler and optional CFG support.

Jun 13, 2025

Minimal release notes: "Automatically generated release for refs/tags/v3.0.2"

Apr 9, 2025