10 best self-hosted transcription tools (2026)
If you want to transcribe audio and video without shipping recordings to a cloud service, the self-hosted world has grown a lot. Here's an honest map of the strongest open-source options in 2026 — what each is actually good at, and where it isn't. We build one of these (Txtify), and we're not going to pretend it's the right pick for everyone.
Quick comparison
| Tool | Best for | Diarization | Translation | Live capture | Setup weight |
|---|---|---|---|---|---|
| Txtify | File/YouTube → subtitles + translation | Not yet | Yes (DeepL) | No | One container |
| Whishper | Subtitle editing UI | No | Yes (LibreTranslate) | No | Multi-service |
| Scriberr | Notes, chat, batch, diarization | Yes | No | No | Medium |
| Speakr | Long meetings + knowledge base | Yes | No | Yes | Medium (alpha) |
| Whisper-WebUI | Feature-rich Whisper front end | Yes (pyannote) | Yes (NLLB/DeepL) | Partial | Heavy (CUDA) |
| Vibe | Offline desktop, many accelerators | Yes | No | Yes | Desktop app |
| Buzz | Desktop, Flathub/Snap/winget | Yes | No | Yes | Desktop app |
| noScribe | Researchers / qualitative work | Yes | No | No | Desktop app |
| OpenTranscribe | Teams needing auth/roles | Yes | No | No | Heavy (enterprise) |
| WhisperLiveKit | Real-time streaming server | Backend-dependent | Streaming | Yes | Server |
Feature sets and release status change quickly — verify against each project's repo before committing.
The rundown
1. Txtify — simplest path from a file or link to subtitles + translation
A single Docker container with a web UI. Paste a YouTube link or upload a file, pick a Whisper model, and export TXT, SRT, VTT, SBV, or PDF — with optional DeepL translation into 30+ languages. Honest limits: no speaker diarization yet, no live capture, and DeepL translation is an external API call. If your job is "video in → translated subtitles out" with minimal setup, it's hard to beat. Try the demo.
2. Whishper — polished subtitle editor
A nice subtitle-editing UI over Faster-Whisper with LibreTranslate. The catch in 2026: the current stable line is frozen while development moves to a v4 rewrite, so treat it as stable but not advancing. Heavier multi-service setup than a single container.
3. Scriberr — notes, chat and diarization
Actively developed, with speaker diarization, batch upload, annotations, a folder watcher and chat via Ollama or OpenAI-compatible models. Broader and more complex than Txtify; no built-in translation.
4. Speakr — meetings and knowledge management
One of the fastest-moving projects: long meeting sessions, diarization, summaries, search, system/mic capture. Still labelled alpha and steadily becoming a full knowledge system with more operational scope.
5. Whisper-WebUI — feature-rich, heavier
File, YouTube and mic input, VAD, background-music separation, pyannote diarization, subtitle formats and NLLB/DeepL translation. Very capable, but CUDA-heavy with a large image and packaged releases that can lag the source.
6. Vibe — strong offline desktop app
Probably the best broad local desktop alternative: batch processing, diarization, live preview, system audio, many exports and NVIDIA/AMD/Intel acceleration. Desktop-first rather than a lightweight self-hosted web service, and it has no translation.
7. Buzz — mature desktop, easy install
Files, video, YouTube, live mic, speaker ID, folder watching, CLI and summaries, distributed through Flathub, Snap and winget. Broad and mature, but desktop-oriented.
8. noScribe — built for researchers
Speaker identification, a transcript editor with audio-following playback, and qualitative-analysis workflows. A strong specialist tool, less suited as a general server service, and moving more slowly than the fastest projects.
9. OpenTranscribe — enterprise-flavoured
Authentication, LDAP/OIDC/MFA, redaction, queues and parallel downloads. The right pick if you need permissions and roles — and heavier to deploy than everything else here (AGPL).
10. WhisperLiveKit — real-time streaming
A real-time local transcription server and web UI with streaming translation and multiple backends. If you specifically need live streaming rather than file-to-subtitle, start here — it's a reminder that live transcription is its own specialist niche.
How to choose
- Subtitles + translation, minimal setup: Txtify.
- Speaker-labelled meeting notes: Scriberr, Speakr, or Whisper-WebUI.
- Desktop app, no server: Vibe or Buzz.
- Live/real-time: WhisperLiveKit or Speakr.
- Auth and roles for a team: OpenTranscribe.
Related: Otter.ai open-source alternative · Whishper vs Scriberr vs Txtify · Generate SRT subtitles · Transcribe YouTube videos · Private, offline transcription