Blueprint by Mozilla.ai on how to transcribe audio files
-
Updated
Jun 13, 2025
Blueprint by Mozilla.ai on how to transcribe audio files
Open-source voice notes to Markdown and Obsidian: record, transcribe and automate watch folders. Self-hosted on Linux, Windows and macOS.
Local-first script-to-audio authoring and rendering for external Speaches text-to-speech servers.
Self-hosted OpenAI-compatible STT/TTS and word alignment. CPU, AMD/Intel Vulkan or NVIDIA CUDA Docker; native Linux, Windows and macOS.
Turn-key Docker stack for self-hosted Open Notebook with host Ollama, Speaches TTS/STT and SurrealDB - chat, embeddings, podcasts, no API key required.
Self-hosted speech-to-text with Whisper, faster-whisper and Speaches. Docker stack with regression-tested model unload fixes and shared GPU lifecycle integration.
Solução desenvolvida para o desafio DIO × Globant — Java & Spring Boot AI Developer.
A flexible Docker Compose setup for running AI tools locally with optional external API support. Choose complete privacy with local models or connect to cloud providers when needed.
To associate your repository with the speaches topic, visit your repo's landing page and select "manage topics."