Jump to content

OpenSpeaks/Tools

Add links
From Meta, a Wikimedia project coordination wiki

The OpenSpeaks tools are offline-first, open-source software built for language documenters, archivists and Wikimedians to document, edit, and archive oral knowledge and language materials. All code is licensed under permissive open-source licenses (MIT or GPL), fully public, and free to use. The design prioritises simplicity, offline capability, and cross-platform access. Internet access is required only for optional upload and API query functions (such as machine translation). The need for these tools within the community was identified during a Rapid Grant-supported project and documented in a research paper.

Timeline: July 2025–June 2027

Tools

OpenSpeaks Subtitler

[edit]
Main article: OpenSpeaks/Tools/Subtitler
OpenSpeaks Subtitler v1.1.5 interface showing translation panels
Purpose

Create, edit, and machine-translate subtitles for oral knowledge and language documentation media, with direct integration to Wikimedia Commons.

Core Features
  1. Load Media and Subtitles
    • Accepts common audio and video formats (MP4, MOV, MP3, WAV).
    • Fetches media files and existing subtitle tracks directly from Wikimedia Commons.
  2. Dummy Subtitle Generation
    • Configurable silence threshold (in dB) and minimum silence length (in milliseconds or seconds).
    • Automatically segments media into draft subtitle chunks.
  3. Timeline and Editing Interface
    • Full timeline view alongside zoomed view of active subtitle lines.
    • Editable text overlay directly on video player with real-time playback updates.
  4. Machine Translation Assistant
    • Side-by-side translation panel for subtitle text.
    • Machine-translation assistance to generate draft translations in multiple target languages.
  5. Language Selection
    • Integrated language database mapping ISO 639 codes to standard names, endonyms, and dialectal varieties.
  6. Export and Wikimedia Upload
    • Saves subtitles as SRT or VTT files with UTF-8 encoding.
    • OAuth authentication to upload completed subtitle tracks directly to TimedText pages on Wikimedia Commons.

Status and Development

  • Stable version published at subtitler.toolforge.org (repository).
  • Development led by Ranjith Siji.
  • Community developers Arpitha Bhandary and Govind Lal T. L. expanded Commons fetching, side-by-side translation, and bug fixes during the Indic Wikimedia Hackathon 2026 in Hyderabad. Opino Gomango contributed early testing and user feedback.
  • Future roadmap: Integration of Automatic Speech Recognition (ASR) draft assistance for supported languages.

OpenSpeaks Bento

[edit]
OpenSpeaks Bento v1.0.0 interface
Purpose

Organise, categorise, tag, batch-rename, inspect, and compress multimedia files (video, audio and image) in local folders using structured naming conventions and archival checks.

Core Features
  1. Organise Tab
    • Local folder selection via File System Access API.
    • Assign editable categories (e.g. A-roll, B-roll) and supplementary tags (e.g. Close-up, Establishing).
    • Built-in preview player for video, audio, and images.
    • Configurable naming convention builder: {language}-{type}-{names}-{subject}_{sequence}.{ext}
    • Batch rename files directly within the local filesystem.
    • Export organisation logs in CSV or TSV format.
  2. Duration Tab
    • Batch calculate individual and total runtimes for audio and video files across folders for project budgeting.
    • Export summary reports in plain text or CSV.
  3. Compress Tab
    • Inspect media properties (duration, resolution, frame rate, audio sample rate, bitrate, codecs).
    • Re-encode files based on user-entered target file sizes or calculated bitrates.
  4. Analyse Media Tab
    • Extract detailed technical metadata including audio/video codecs, resolution, loudness metrics, and SHA-256 checksums for archival verification.
Status and Tech Stack
  • Stable version 1.0 released on 1 July 2026 at bento.toolforge.org (repository).
  • Self-contained single-page utility working fully offline in Chromium-based browsers after initial load.
  • UI built using Wikimedia's Codex design system, ensuring WCAG accessibility compliance.

OpenSpeaks Tome

[edit]
OpenSpeaks Tome v1.5.1 interface
Purpose

Generates structured wikitext, templates, licensing headers, and metadata records for uploading audio, video, image, and document files to Wikimedia Commons. Designed around the OpenSpeaks Oral History Framework, it provides interoperability between Wikimedia Commons and external language archives.

Core Features
  1. Structured Forms across Five Tabs
    • Basic Info: File title, original title, description, language, recording location, dates, runtime, and source.
    • Production: Project name, speaker/cast names, genre, topic, producer, director, and external identifiers (DOI, EIDR, IMDb, ISRC).
    • People and Attribution: Dedicated tab for interviewees, interviewers, transcribers, and reviewers. Captures full names, Wikidata QIDs, birth year, gender, and roles. Generates live preview of author and moral rights lines.
    • Subtitles and Oral History: Captures TimedText page references, subtitle creators, recording context, equipment, consent status (verbal, written, informed, revocable), sensitive content flags, and community review notes.
    • License and Categories: License selector for Commons-compatible licenses and multi-line category input.
  2. Integrated Language Database
    • Autocomplete language selector mapping ISO 639 codes to standard names, endonyms, and dialectal varieties.
    • Cross-checked against Glottolog and Wikipedia datasets.
  3. Metadata Import
    • OpenSpeaks JSON v1.2 schema files.
    • Existing Commons wikitext (parses {{Information}} and {{OpenSpeaks}} templates).
    • Endangered Languages Archive (ELAR) session metadata (.opex / IMDI XML from Lameta).
    • Language Archives Cologne (LAC) metadata (BLAM / CMDI format).
  4. Preview and Export
    • Output formats: Wikimedia wikitext (with {{Information}} and nested {{OpenSpeaks}} templates), OpenSpeaks JSON v1.2, Markdown, and HTML.
    • One-click clipboard copy and file downloading (.txt, .json, .md, .html).

Status and Development

  • Stable version published at tome.toolforge.org (repository). Formerly titled Wikimedia Commons Metadata Generator.
  • Mentored by Subhashish Panigrahi and Jnanaranjan Sahu during the Indic Wikimedia Hackathon 2026. Features developed by Arpitha Bhandary and Govind Lal T. L., with editor support from Bharathesha A.
  • React single-page application hosted on Toolforge.

OpenSpeaks Media Dashboard

[edit]
OpenSpeaks Media Dashboard v1.2.0 interface
Purpose

Enables language archivists and Wikimedians to monitor, measure, and report the usage and readership impact of media files contributed to Wikimedia Commons.

Core Features
  1. Category-Based Impact Querying
    • Queries Wikimedia public APIs in real time using any Wikimedia Commons category containing audio, video, or images.
  2. Metrics and Visualisations
    • Tracks which Wikimedia pages embed files from the specified category.
    • Displays monthly pageview statistics, breakdowns by Wikimedia project, breakdowns by language edition, and file-type distributions.
    • Generates ranked tables of top-viewed pages.
  3. Export and Reporting Options
    • Export charts as SVG or PNG images.
    • Export raw query datasets as CSV files.
    • Generates copy-ready wikitext analysis reports for publishing documentation directly on wiki pages.
    • Generates shareable report links and QR codes.

Status and Tech Stack

  • Released as a stable tool at omd.toolforge.org (repository).
  • Client-side web application querying live APIs directly without external database storage. Includes built-in user documentation tabs.

Hackathons and Community Mentorship

[edit]

During 25–28 June 2026, OpenSpeaks participated in the Indic Wikimedia Hackathon 2026 at IIIT Hyderabad. Subhashish Panigrahi and Jnanaranjan Sahu co-mentored community developers Arpitha Bhandary and Govind Lal T. L., who built core functionality and resolved bugs across Subtitler and Tome. Wikimedian Bharathesha A provided input for the user workflow, while Ranjith Siji managed code reviews and pull request integration. OpenSpeaks Fellow Opino Gomango provided field-testing input.

Needs

[edit]
  • Community testing across Subtitler, Tome, Bento, and Media Dashboard with real field recordings and Commons categories.
  • Developer support for Automatic Speech Recognition (ASR) integration in Subtitler and deeper Wikidata entity linking in Tome.
  • Ongoing feedback from language documenters on workflow integration via the Meta-wiki talk page.

Blogs, reports & publications

[edit]

Inspirations

[edit]
  • LinguaLibre Dictionary for creating a dictionary using recordings on Commons: Excellent way to organise community-based recording sessions for those interested in building a dictionary
Retrieved from "https://meta.wikimedia.org/w/index.php?title=OpenSpeaks/Tools&oldid=31120098"