Transcriber

by Sébastien Dubois
5
4
3
2
1
Score: 54/100

Description

Transcriber converts images in your Obsidian vault to Markdown using Ollama vision models running locally on your machine.

Reviews

No reviews yet.

Stats

25
stars
4,378
downloads
4
forks
190
days
2
days
2
days
14
total PRs
0
open PRs
9
closed PRs
5
merged PRs
1
total issues
0
open issues
1
closed issues
94
commits

Latest Version

3 days ago

Changelog

2.2.0 (2026-11-17)

You can now choose where the text of a transcription goes. The new settings are in Settings, Transcriber, under Transcription settings. They are all empty or off by default, so nothing changes until you set one. Thanks to bobhalford for the request.

Put transcription files in one folder

With a large media folder, the .md files got lost among the images. Set Output folder and every transcription file goes to that folder, whatever folder the image is in. The folder is created if it does not exist. Leave it empty and each file stays next to its image, as before.

File name prefix adds text in front of the name of each transcription file: OCR_ gives OCR_photo.md. It works with and without an output folder, and it cannot contain \ / : * ? " < > |.

One thing to know: with an output folder, two images that have the same name (in two different folders) share one transcription file. The second one is skipped, or replaces the first when Overwrite existing files is on.

Insert the transcription into the current note

Turn on Insert into the current note and the text goes into your note. No separate file is created.

  • Transcribe all images in current note: each transcription goes in a new paragraph right after its image, in the order of the note.
  • Transcribe this image (right-click an image embed in the editor): the text goes after the image.
  • Transcribe image (right-click an image in the file explorer): the text goes into the note you have open. It lands after the image if the note embeds it, at the cursor if it does not. Reading view has no cursor, so there the text goes at the end of the note.

If no note is open (for example, the image itself is the active tab), a file is created as usual. Folder transcriptions always create files. In this mode, the frontmatter tags are not added and an existing transcription file does not stop the transcription, so running the action twice inserts the text twice.

The notice says where the text went

When a single image is done, the notice now shows the path of the transcription file, or the note the text was inserted into. It stays on screen for 10 seconds. It also tells you when an image was skipped because its transcription already exists (Overwrite existing files is off).

Documentation

The documentation now recommends glm-ocr:latest as the default model.


Enjoying this plugin?

README file from

Github

Transcriber

Transcribe images in your vault to Markdown using local Ollama vision models. Point it at any image and get structured Markdown back — headings, lists, tables, code blocks — all extracted by a vision AI running on your own machine. No data leaves your computer.

What it does

  • Transcribe a single image via the command palette or right-click context menu
  • Batch-transcribe an entire folder of images (with optional subfolder inclusion)
  • Creates a .md file alongside each image with the transcribed content, or in one output folder of your choice, with an optional file name prefix
  • Or inserts the text into your note, right after the image, without a separate file
  • Install, select, and remove AI models directly from the command palette — no terminal needed
  • Progress tracking for batch operations with per-file status
  • Configurable prompt so you can tailor the transcription instructions
  • What's new after updates. After a plugin update, a one-time dialog shows the release notes you just received (including skipped versions) with ways to support development. Never shown on fresh installs or regular restarts.

The plugin recommends these vision models for transcription:

glm-ocr:latest, maternion/LightOnOCR-2:1b, qwen3.5:2b, qwen3.5:4b, qwen3.5:9b, qwen3.5:27b, qwen3.5:35b

Any other Ollama vision model can be installed directly from the settings or via the Ollama CLI.

Prerequisites

  • Ollama installed and running locally
  • Desktop Obsidian (this plugin is desktop-only)

Installation

  1. In Obsidian, go to Settings → Community plugins.
  2. Disable Restricted mode if it's enabled.
  3. Select Browse, search for Transcriber, install it, then enable it.

You can also browse the catalog on the Obsidian Community website.

Manual installation

If the plugin isn't listed in the community catalog yet (or you want a specific version):

  1. Download main.js, manifest.json, and styles.css from the latest release.
  2. Copy them into <Vault>/.obsidian/plugins/image-transcriber/.
  3. Reload Obsidian and enable Transcriber in Settings → Community plugins.

BRAT (bleeding edge)

BRAT (Beta Reviewers Auto-update Tool) installs plugins straight from a GitHub repo and keeps them updated automatically. Use this if you want the latest commits — things might break.

  1. Install Obsidian42 - BRAT from Settings → Community plugins → Browse and enable it.
  2. Run BRAT: Add a beta plugin for testing from the command palette.
  3. Paste https://github.com/dsebastien/obsidian-transcriber.
  4. Select the latest version and confirm.
  5. Enable Transcriber in Settings → Community plugins.

Getting started

  1. Install the plugin (see Installation above).
  2. Enable it
  3. Open Settings > Transcriber and verify the Ollama server URL (default: http://localhost:11434)
  4. Click Test to confirm the connection
  5. Install a model: open the command palette (Ctrl/Cmd+P) and run Install AI model, or install from settings
  6. Right-click any image in your vault and select Transcribe image

Documentation

See the user guide for detailed usage, configuration, and troubleshooting.

License

MIT

My other Obsidian plugins

Plugin What it does
Agentic Resource Discovery Server Local-first Agentic Resource Discovery publisher and registry that serves your AI skills and tools to agents over a local HTTP and MCP server
Book Exporter Export books (one manifest note + linked chapter notes) to EPUB and PDF via Pandoc
Bookshelf Base Display your notes as a visual bookshelf via a custom Bases view
Dataview Serializer Serialize Dataview queries to Markdown, and keep the Markdown representation up to date
Expander Replace variables across your vault using HTML comment markers. Supports static values and dynamic functions
Ghost Publish Publish your vault notes to a Ghost blog with configurable presets for tags, newsletters, and frontmatter conventions
Graph Explorer Base View A custom Bases view that renders notes as an interactive force-directed graph with explored/unexplored tracking
Hidden Folders Access Index hidden root-level folders (e.g. .claude) so they appear in the file tree, metadata cache, and Bases
Journal Bases Custom Base views for journaling and periodic reviews
Kanban Action Planner Render your notes as configurable Kanban boards and calendars inside Bases, with statuses, ordering, relationships, and scheduling
Life Tracker Capture and visualize the data that matters in your life
Note Village A 2D pixel art village where your notes become villagers you can explore and chat with using AI
Obsidian Starter Kit Adds strong typing support and powerful automation support for notes
Remarkable Synchronizer Connect to the reMarkable cloud, list, download, and sync notebook pages as images
Replicate Use AI models with ease via the Replicate.com integration
REST and MCP server Exposes CLI commands as RESTful API endpoints and an MCP server for AI tool integration
Time Machine Browse, compare, and restore previous versions of your notes using built-in file-recovery snapshots
Typefully Publish social media posts with ease using the Typefully integration
Update Time Automatically update front matter to include creation and last update times

Everything I build is documented in my newsletter and on my YouTube channel.

News & support

To stay up to date about this plugin, Obsidian in general, personal knowledge management and note-taking:

If this plugin is useful to you, here are the best ways to support my work ❤️:

Found a bug or have an idea? Open an issue.