Speakr - Self-Hosted AI Transcription and Intelligent Note-Taking Platform. Alternative to Otter.ai

[Teiluj]

Speakr - Self-Hosted AI Transcription and Intelligent Note-Taking Platform, alternative to Otter.ai

---

• Main Page: https://murtaza-nasir.github.io/speakr/
• Git: https://github.com/murtaza-nasir/speakr
• Licence: Dual-licensed — AGPL-3.0 or Commercial License (choose one)
• Dockerfile: Yes (used by the Docker Compose setup)
• Docker-compose: Yes - Recommended setup via docker compose up -d with an env template
• Demo: No public demo available

---

• Summary:
  • Self-hosted transcription platform that captures audio (mic, system/tab audio, or uploaded files), transcribes via bring-your-own ASR engines (self-hosted WhisperX, OpenAI, AssemblyAI, and more), and turns recordings into searchable knowledge with speaker diarization, summaries, and event extraction
  • Features per-recording AI chat, "Inquire Mode" agentic semantic search across your entire library, smart tags with custom prompts, retention policies, and automated exports
  • Multi-user with OIDC SSO, groups with auto-shared recordings, REST API with Swagger UI, signed webhooks, and usage budgets — packaged as a mobile-first PWA in seven languages

---

• Notes:
  • Perfect for privacy-conscious users and teams wanting to self-host the full pipeline (e.g. WhisperX + local LLM) so recordings and transcripts never leave your environment
  • Requires external ASR and LLM API services (or self-hosted companions like whisperx-asr-service) to function — it orchestrates transcription rather than including a built-in engine

---

• Alternative to / Libhunt link: https://alternativeto.net/software/otter-ai/

• Screenshots:
  Available here: https://murtaza-nasir.github.io/speakr/screenshots/

  • The main view: recordings list, transcript, and summary
    

  • Video playback synced to the transcript
    

  • Audio player with the generated summary
    

  • Per-recording stats and speaker breakdown
    

[timconsidine]

@Teiluj I will take a look at this (no promises).

Without hijacking the thread, have you seen my LocalPlaud community app ?
It’s not the same but it’s in the same space, and it is already packaged / published.
You don’t need a Plaud device to use LocalPlaud - now copes with dictating into it and uploading recordings.

[Teiluj]

Hi @timconsidine

@Teiluj I will take a look at this (no promises).

Even for considering, thank you already

Without hijacking the thread, have you seen my LocalPlaud community app ?
It’s not the same but it’s in the same space, and it is already packaged / published.

I actually did have a look and was quite interested but I am not familiar with Plaud and by the looks of it, it is the kind of application that I would initially ran away (maybe wrongly) from because of privacy and security concerns.

You don’t need a Plaud device to use LocalPlaud - now copes with dictating into it and uploading recordings.

I did not know this and it goes miles into alleviating my Plaud concerns. I will git it a try!

[timconsidine]

@Teiluj understood about privacy.

It’s actually why I built it, so my Plaud recordings don’t get processed on Plaud, just pass through.

But then I found that it’s just as easy to dictate into the interface or upload a recording (input device agnostic).
So actually I’m using my Plaud less and going direct.

One thing I like about LocalPlaud is the ability to make multiple small recordings and a composite summary is auto generated. I use this feature as a Mood Journal. Dictate snippets. Have them summarised.

[jdaviescoates]

@Teiluj said:

Speakr - Self-Hosted AI Transcription and Intelligent Note-Taking Platform, alternative to Otter.ai

This sounds great! Thanks for sharing.

[timconsidine]

I did a quick research pass :

• viable
• defaults to full PyTorch which makes it a fat image, but could try the lite version with some compromises
• latest release is v0.10.5-alpha so still young as a repo and all recent releases are alpha

None of those are blockers, just issues to deal with, might be a bumpy road until it is stable.

I will add it to my 'Consider' list, maybe watch the repo a little, but if anyone else wants to get there first, please go ahead, don't let my looking at it dissuade any packagers.