Synthid-Bypass Releases

A maintenance release focusing on CLI improvements, including the ability to create an admin with predefined credentials, alongside dependency updates and minor icon fixes.

Sep 17, 2025

A feature release adding padding options to multiple CLI commands, allowing for better UI control similar to CSS values.

Sep 5, 2025

A feature release introducing optimizations for the pipeline form and updates to Bitnami Helm charts and images.

Sep 2, 2025

An update to the Ultravox speech model introducing new variants and evaluation benchmarks, with improvements in handling noisy audio.

Aug 18, 2025

A bug fix release addressing database service operations by updating the method used to handle data.

Aug 13, 2025

A major release featuring a comprehensive V3 refactoring, migration to Prisma, implementation of RBAC, and the addition of numerous new templates and internationalization support.

Aug 11, 2025

A bug fix release addressing specific issues with environment variables and UI sizing in the Gum CLI tool.

Jun 23, 2025

A maintenance release focused on stability and CI/CD improvements, featuring fixes for timeout error detection, linting issues, and documentation updates regarding editor usage.

May 30, 2025

This release adds a visual indicator for item positions on the help screen and addresses several bugs across the choose, confirm, spin, table, and filter commands to improve shell completion and functionality.

Mar 11, 2025

This release introduces seven new application templates (Piwigo, searxng, Kataeomon, CaddyGen, Airbroke, Outline, and Trailbase) and enhances GitHub integration with modified views and BaseUrl usage checks. It also adds version fields for addons and visibility of configured addons in the templates list.

Feb 26, 2025

This release brings significant performance improvements to transcription accuracy, speech-to-question answering, and translation metrics, alongside a massive expansion of language support from 15 to 42 languages. The weights have been pushed to Hugging Face, making v0.5 the new default for Realtime APIs.

Feb 11, 2025

A minor patch release that resolves high memory and CPU usage issues encountered when using soft-wrap functionality in the pager command.

Jan 28, 2025

Upgrades the Whisper encoder to Whisper-large-v3-turbo and adds support for six new languages, bringing the total supported languages to 15. Includes increased English training data and significant improvements to the training pipeline, including FSDP support for the 70B model.

Nov 12, 2024

Upgrades the Whisper encoder from Whisper-small to Whisper-medium and expands multilingual support to include 9 languages. The model is trained on a larger multilingual dataset and shows significant improvements in BLEU scores for speech translation tasks.

Aug 27, 2024

Introduces improved speech understanding using a frozen Llama 3.1 8B core. Key changes include the introduction of Knowledge Distillation loss calculation, synthetic data augmentation, and the addition of a Gradio demo for easier testing.

Aug 23, 2024