DeLive

DeLive is a free, open source desktop transcription workspace for Windows, macOS and Linux. It captures system audio directly, so meetings, lectures and podcasts can be transcribed without a microphone, and it also transcribes uploaded audio and video files. Transcripts are then reviewed with AI for correction, summaries, chat and mind maps, with data stored locally. It is aimed at students, researchers, journalists and professionals who want flexible speech recognition without locking into one provider. DeLive supports twelve speech recognition backends across streaming, batch and local modes: Soniox, Volcengine, ElevenLabs, Mistral AI, Gladia, Deepgram, AssemblyAI, Cloudflare, SiliconFlow, Groq, any OpenAI compatible service, and whisper.cpp running on the device. Live transcription covers more than 60 languages with speaker detection, word level timestamps and translation. The AI Review Desk produces summaries, action items, keywords and chapters, supports multi thread chat with a transcript, and generates interactive mind maps with Markmap. AI Correction offers a Quick Fix mode and a Review and Fix mode with a side by side diff. An always on top floating caption window shows source, translated or bilingual subtitles, and developers get a REST API, WebSocket streaming and an MCP server that works with Claude Desktop and Cursor. Backups can go to WebDAV or S3 compatible storage. DeLive is completely free and released under the Apache 2.0 licence. Costs arise only from the cloud speech recognition or AI model providers a user chooses to connect. Local storage uses IndexedDB, with credentials encrypted through Electron safeStorage and context isolation enabled.