SourceQuote — Product Overview & Marketing Reference

Overview

SourceQuote is an open-source audio transcription workspace designed for journalists, researchers, podcasters, and anyone who works seriously with recorded speech. It combines AI-powered transcription, interactive waveform editing, speaker management, and flexible sharing tools into a single, cohesive platform.

Unlike most transcription tools, SourceQuote treats the transcript not as a final output but as a living document — one that stays synchronized with its source audio, can be precisely edited, annotated, and shared in ways that preserve the connection between the written word and the human voice behind it.

SourceQuote is available as a free, open-source desktop application, as a self-hosted server deployment, and as a managed cloud service at sourcequote.org.

Origin Story

SourceQuote started not in a newsroom or a podcast studio, but at a tabletop RPG table.

Late in 2024, the developer's D&D group began taking their campaign seriously enough to want to document it — key story beats, character moments, specific dialogue — with the goal of eventually creating content from it. Sessions were recorded on a phone using Google Recorder. It didn't go well. Five-hour sessions drained the battery, the built-in mic captured everything except clear speech, and one session stopped recording mid-way through and was lost entirely.

Switching to a proper microphone solved the audio quality problem, but created a new one: Google Recorder only transcribes audio it captures live. Pre-recorded files weren't supported. So a Python script using OpenAI's Whisper model was written to handle transcription locally — free, high quality, and no data leaving the machine.

That solved transcription. But working with a five-hour transcript turned out to be its own problem. Searching across hours of dialogue, correcting speaker labels, syncing the text back to the audio to verify specific quotes, sharing the results with the rest of the group — no existing tool did all of that. So one was built.

An early version of that tool was shown to a friend who works as a journalist. They immediately began listing every frustration they had with their current transcription workflow, and SourceQuote addressed most of them on the spot. That was the moment it became clear this was worth building properly, for a broader audience.

The result is SourceQuote.

Development Philosophy

SourceQuote was built around a frustration familiar to anyone who works with recorded audio professionally: the tools available are either too expensive, too limited, too closed, or too disconnected from the actual audio they represent. Transcription services produce text files. Waveform editors do not understand language. Quote management is done manually in separate documents.

SourceQuote was developed to solve this by treating the transcript and the audio as inseparable — every word in the transcript is anchored to its exact position in the audio, and every action in the interface (editing, playback, navigation, export) respects that relationship.

Development Process

Phase 1: Prototype (Pre-March 2026)

SourceQuote began as a back-and-forth conversation with an AI assistant. The initial prototype was built iteratively through Claude (claude.ai), prompting for features, getting code back, testing it, and asking for adjustments — the way many solo developers now bootstrap an idea quickly without a team.

The result was a functional but structurally chaotic codebase: a largely monolithic file where features were piled on top of each other, concerns were mixed, and the code was difficult to extend or maintain. It worked, but it was fragile. Showing it to a journalist friend made clear it had real potential — and that potential required a proper foundation.

Phase 2: The Untangling (March 2026 — V0.1.0)

The first step after recognizing the project's potential was a deliberate structural overhaul. The monolithic prototype was decomposed into a proper application architecture: separate modules for the backend API, database layer, frontend panels, and utilities. A CI/CD pipeline was established using GitLab, and the first formal unit and integration test suite was added.

This was not glamorous work, but it was necessary. The V0.1.0 release (March 11, 2026) represented the prototype's ideas expressed in a codebase that could actually grow.

Over the following weeks, the pace of structured development was rapid:

• Export functionality (PDF, DOCX, CSV, plain text) added
• Full-text search across transcripts implemented
• Multiple audio format support (MP3, WAV, FLAC, OGG, M4A, AAC, WebM, Opus)
• JSDoc documentation system established
• Light and dark themes introduced
• Server deployment mode made production-ready with deployment tutorials

Phase 3: Infrastructure & Multi-User (Late March 2026)

The application's data layer was overhauled from a flat file system to a PostgreSQL relational database — a significant architectural change that enabled proper multi-user support. Firebase authentication was integrated, introducing user accounts, project ownership, and folder-level sharing permissions.

The standalone desktop executable was packaged and released, along with the project's open-source license (AGPL 3.0 with Commons Clause). A first pass at security hardening addressed SQL injection vectors and iframe content security policies.

By the end of March, SourceQuote was a multi-mode application: local desktop, self-hosted server, and the beginnings of a cloud deployment path.

Phase 4: Sharing & Transcription Depth (April 2026 — V0.3)

The V0.3 release added the capabilities that make SourceQuote useful for publishing and professional review workflows:

• Word-level transcript data — individual words within segments became addressable, enabling word-accurate splitting and Live Quote generation
• Usage analytics and admin dashboard — page-level tracking and reporting for server deployments

Phase 5: Annotation, Audio Processing & Productization (April–May 2026 — V0.4)

V0.4 was the most feature-dense release, adding the capabilities that distinguish SourceQuote from simpler transcription tools and completing the groundwork for a paid product:

• Transcript sections — structural navigation breaks across long recordings, visualized both in the transcript panel and the waveform
• Annotations and notes — segment-level notes attachable to specific timestamps
• Audio effect pipeline — muting, gain adjustment, and effect chains applied to audio segments, with waveform visualization of effects
• Off-the-record segments — ability to mark sections of a recording as confidential, redacted from exports and presentations
• Parallel transcription — multiple transcription jobs run concurrently in the background; users can navigate away and return when complete
• Project versioning — full version history with restore capability
• Stripe payment integration — full subscription tier system with checkout, customer portal, webhooks, and usage enforcement
• Desktop installer — Windows installer with bundled binaries, replacing the earlier standalone script approach; GPU-accelerated (CUDA) and CPU-only builds

The app was also formally renamed from its working title Waveform Studio to SourceQuote in this release, with updated branding, logo, and front-facing copy.

Phase 6: Cloud Deployment & Production Readiness (May–June 2026 — V0.5)

V0.5 focused on cloud infrastructure, sharing improvements, and hardening the application for public launch:

• Google Cloud deployment — first successful end-to-end cloud deployment (Cloud Run, CloudSQL, GCS, Modal for cloud transcription)
• Mobile-responsive layouts — front pages and workspace optimized for mobile viewports
• Collapsible panels and UI polish — workspace panels made collapsible for focused editing; tooltips standardized across the application
• Sharing overhaul — presentation mode updated with playlist support, and link-based access controls refined
• Security audit — a comprehensive pass surfacing and fixing XSS vulnerabilities (innerHTML usage), SSRF bypass vectors, privilege escalation flaws, injection risks in SQLite and Stripe callbacks, auth token exposure in media URLs, missing MIME validation, and rate limiting gaps across multiple endpoints
• Privacy policy and terms of service — added with account signup acknowledgment flow
• Email confirmation system — verified email addresses required for new accounts
• Production/development environment separation — .env file separation, DEPLOY_MODE switching, Stripe sandbox-to-live transition
• Dependency audit — pip-audit integrated into CI pipeline to catch vulnerable packages
• SEO and discoverability — conditional metadata and structured markup for the cloud deployment
• Coupon/discount code system — Stripe coupon integration for promotional pricing

Architecture diagrams documenting the system's three deployment modes were also created during this phase.

SourceQuote launched publicly at sourcequote.org in June 2026.

Open Source First

SourceQuote is fully open source under the AGPL 3.0 license with Commons Clause. This means:

• The full source code is publicly available and auditable
• Anyone can download, run, modify, and self-host the application at no cost
• The Commons Clause restricts selling SourceQuote itself as a commercial product without permission, protecting the project from being repackaged and resold without contribution back to the ecosystem

The only paid component is the managed SaaS deployment at sourcequote.org — a hosted version of the same open-source software that removes the need for users to set up and maintain their own server infrastructure. Subscription revenue primarily covers the cost of backend cloud services (compute, storage, transcription), with a modest markup to support ongoing development. One-time donations are also welcomed and appreciated.

This model means that any organization or individual with technical capacity can run SourceQuote entirely for free, with no feature limitations, no licensing fees, and no dependency on any external commercial service.

Deployment Modes

SourceQuote offers three distinct ways to run the application, each suited to different needs and technical environments.

1. Local Desktop Application

SourceQuote ships as a standalone desktop executable for Windows (Linux and macOS builds are planned and require only compilation on those platforms). The desktop app runs entirely offline — no account, no internet connection, no cloud dependency required.

All data is stored locally on the user's machine (SQLite database and project files in ~/Documents/SourceQuote/). Transcription is performed using local AI models (OpenAI Whisper via faster-whisper), optionally accelerated by a CUDA-compatible GPU.

This mode is ideal for users with privacy requirements, unreliable internet access, or those who simply prefer to keep their data entirely under their own control.

2. Self-Hosted Server

Organizations can deploy SourceQuote on their own infrastructure — on-premises or on a private cloud — using the provided server deployment configuration. This mode supports multiple users, authentication (via Firebase or custom configuration), subscription management, and all cloud features, while keeping all data within the organization's own environment.

Self-hosting requires technical setup (Flask web server, PostgreSQL database, optional cloud storage), but detailed configuration documentation is provided. The software is identical to the managed cloud version.

3. Managed Cloud (sourcequote.org)

For users who want the full SourceQuote feature set without infrastructure overhead, the managed cloud deployment at sourcequote.org provides immediate access. Users create an account, upload audio, and start working — all storage, compute, and maintenance is handled transparently.

Subscription tiers on the managed service determine storage limits, monthly transcription hours, and access to larger (more accurate) Whisper AI models.

Core Capabilities

AI Transcription

SourceQuote uses OpenAI Whisper — one of the most accurate open-source speech recognition models available — to transcribe audio automatically. Transcription supports over 99 languages and can handle a wide range of audio formats including MP3, WAV, FLAC, OGG, M4A, AAC, WebM, and Opus.

Users can select from multiple Whisper model sizes depending on the balance they want between speed and accuracy. Smaller models transcribe faster; larger models produce more accurate results, especially for technical vocabulary, accents, and lower-quality recordings.

Transcription can be performed locally (using the user's own hardware) or via cloud-based processing (in the managed service). Individual segments of an existing transcript can also be selectively re-transcribed without reprocessing the entire file.

Speaker Diarization

SourceQuote automatically identifies and separates different speakers in a recording using Pyannote, an industry-grade speaker diarization library. Each speaker is assigned a label, a display name, and a color for visual organization throughout the transcript.

For improved accuracy, users can upload voice sample files for known speakers. The system uses these samples to better match detected speakers to real people, reducing errors in multi-person recordings.

Waveform-Transcript Synchronization

The core differentiator of SourceQuote is its interactive workspace, where an audio waveform and the corresponding transcript are displayed simultaneously and kept in continuous sync.

Clicking any word or segment in the transcript immediately jumps playback to that point in the audio. Playing the audio highlights the corresponding text in real time. This bidirectional synchronization means users never lose their place, and every edit they make to the transcript can be immediately verified against the original audio.

The waveform display supports zoom, scroll, and segment-level selection, giving users precise control over both listening and editing.

Transcript Editing

SourceQuote provides a full suite of transcript editing tools:

• Text editing — correct transcription errors directly in the transcript
• Segment splitting and merging — restructure how the transcript is broken into units
• Timing adjustment — fine-tune the start and end times of segments
• Speaker reassignment — correct speaker labels for any segment
• Undo/redo — full edit history for confident experimentation
• Annotations and notes — attach notes to specific segments for reference or collaboration

All edits remain anchored to the source audio — there is no way to accidentally detach the transcript from its timestamps.

Audio Redaction & Processing

SourceQuote includes a full audio effect pipeline that operates at the segment and word level, with changes rendered directly into the audio served to viewers and listeners.

• Word-range muting — select any span of words in the transcript to mute them; muted text is struck through in the editor and the corresponding audio is silenced in all shared and exported versions
• Off The Record paragraphs — mark entire paragraphs as OTR; they appear only as a redaction placeholder to viewers and collaborators, with text and audio stripped server-side before delivery
• Voice transforms — AI-based modification of speaker voices for individual segments, including pitch shifting and anonymization for contexts where speaker identity must be protected
• Effect chains — apply other audio processing effects (gain adjustment and others) to specific segments, prerendered into the served audio

Once any redaction or effect is applied, the original audio behind those sections cannot be accessed by anyone other than the editor — not viewers with shared access, not public presentation audiences. This guarantee is enforced at the point of audio delivery, not as a permission layer that could be circumvented.

Live Quotes

Live Quotes are one of SourceQuote's most distinctive features, designed specifically for journalists and publishers who embed quoted material in digital articles and web pages.

A Live Quote is a self-contained, embeddable audio-text snippet generated from any segment of a transcript. It displays the quoted text alongside a playback control, allowing readers to hear the speaker say the exact words they are reading — simultaneously, in context, with the speaker's real voice and tone.

Live Quotes can be embedded into any webpage using a standard embed snippet (similar to embedding a YouTube video or a tweet), or opened directly in a browser as a standalone page. This makes them usable in any CMS, blog platform, or news site without requiring any special integration.

For journalism specifically, Live Quotes serve as a form of primary source verification — readers can confirm not only what was said but how it was said, reducing the possibility of misquotation and increasing trust in published content.

Export & Sharing

SourceQuote supports multiple export formats to fit different publishing workflows:

• Microsoft Word (.docx) — formatted transcript document suitable for editorial review
• PDF — print-ready transcript with speaker labels and timestamps
• CSV — structured data export for analysis or import into other tools
• Plain text — simple, portable transcript without formatting

In addition to file export, SourceQuote supports:

• Presentation mode — a clean, shareable read-only view of the transcript, accessible via a link, formatted for comfortable reading
• Public/private sharing — control who can view or edit projects via link-based access controls
• Live Quote embeds — audio-anchored quote snippets for web embedding (described above)

Project Organization

SourceQuote includes a full project management system for users managing multiple recordings:

• Folders — organize projects into hierarchical folder structures
• Folder and project-level sharing — grant specific users viewer or editor access to individual projects or entire folders
• Version history — restore any previous state of a transcript
• Project duplication — use existing projects as templates for new ones
• Project compression — reduce storage footprint of completed projects

Customization & Appearance

The SourceQuote interface supports:

• Light, dark, and system-adaptive themes
• Multiple built-in color schemes with full custom color picker support
• Configurable keyboard shortcuts
• In-app tutorial and onboarding for new users
• Tooltips and contextual help throughout the interface

Subscription Tiers (Managed Cloud)

The following tiers apply to the managed cloud service at sourcequote.org. Self-hosted and local deployments have no tier restrictions.

Annual plans offer a 20% discount versus monthly billing. Subscription management, billing history, and plan changes are handled via the Stripe customer portal.

Why Use SourceQuote?

For Journalists & Researchers

• Transcribe interviews accurately, with speaker attribution, in minutes rather than hours
• Verify every quote against the original audio without leaving the app
• Embed Live Quotes directly into published articles so readers can hear sources in their own voice
• Export to Word for seamless integration with existing editorial workflows
• Keep sensitive recordings entirely on your own hardware with the local desktop app

For Podcasters & Audio Producers

• Generate accurate transcripts for show notes, accessibility, and SEO
• Use the waveform editor to identify and annotate key moments
• Share formatted, readable transcripts with listeners or collaborators
• Organize multi-episode projects in folders with consistent speaker profiles

For Privacy-Conscious Users & Organizations

• Fully open-source code — audit everything, trust nothing blindly
• Self-host on your own infrastructure with no external data dependencies
• Run entirely offline on the desktop app with no network connectivity required
• AGPL license ensures any modifications to the software remain open

For Teams

• Multi-user project access with role-based permissions (viewer, editor, owner)
• Folder-level sharing for organized collaborative workflows
• Admin dashboard for user and usage management (server/cloud deployments)
• Version history to track and restore changes across collaborators

Audience Pitches

The following are marketing pitches tailored to each primary user segment. They are intended as source material for landing pages, social media copy, outreach, and promotional content.

Journalists & Investigative Reporters

You recorded the interview. Now you need the quote — the exact one, with the exact inflection, attributed to the right person, ready for publication.

Most transcription tools stop at the text file. SourceQuote goes further. Every word in your transcript is locked to its precise moment in the audio. Click a sentence and the audio jumps there. Play the audio and the transcript follows along. You never have to scrub through a recording hunting for a line — you just read, and verify, and move on.

Speaker diarization automatically separates voices from the moment the file is uploaded. Assign names, correct misattributions, and end up with a transcript where every quote is already sourced. And when it is time to publish, Live Quotes let you embed a snippet directly into your article — text and audio together, so readers can hear your source in their own voice, in context, without leaving the page.

Source protection is built into the editing workflow, not bolted on afterward. Paragraphs can be marked Off The Record — they appear as a redacted placeholder in any shared or published view, and their audio is stripped server-side before it ever reaches a reader. Individual word ranges can be muted directly in the transcript, striking them from the text and removing them from the audio in all shared versions. Once these restrictions are applied, the original audio cannot be accessed by anyone but the editor — not viewers, not collaborators with shared access, not the published presentation. What the editor marks private stays private, by design.

For sensitive recordings, SourceQuote's desktop app requires no internet connection and stores nothing outside your own machine. For teams and newsrooms, the self-hosted deployment keeps your data on your own infrastructure. No third-party servers. No upload risk. No subscription required to access the source code.

SourceQuote is built for the work that happens between the recording and the byline.

Researchers & Academics

Interview data is only as useful as your ability to work with it. Hours of recorded conversation — structured or unstructured, one speaker or many — have to become searchable, citable, and shareable. That process is traditionally slow, error-prone, and disconnected: transcription in one tool, editing in another, annotation in a third.

SourceQuote brings it together. Transcribe recordings automatically with AI, then move through the transcript with the audio always one click away. Search across the full text of a project to find every instance of a term, theme, or name. Split and merge segments to match your analytical framework. Attach notes directly to specific moments in the recording for qualitative coding.

Multiple audio formats are supported, so it does not matter whether your recordings come from a field recorder, a phone, a Zoom call, or archival sources. And because SourceQuote is open source and can run entirely offline, it is appropriate for sensitive participant data and IRB-compliant research contexts.

Participant confidentiality is enforceable at the content level, not just the access level. Specific word ranges can be muted — struck from the transcript and removed from the audio in all shared or exported versions. Entire paragraphs can be marked Off The Record, appearing only as a redaction placeholder to anyone outside the editor role. Once applied, these restrictions are enforced server-side: the original audio behind a muted or OTR section cannot be accessed by viewers or collaborators, regardless of their permission level. Speaker voice anonymization is also available for contexts where even the voice itself must be protected.

Export transcripts to CSV for quantitative analysis, to Word for writing integration, or share a read-only presentation link with collaborators or supervisors. Organize years of fieldwork into folders. Maintain version history so no edit is ever permanent.

Research produces data. SourceQuote helps you actually use it.

Podcasters & Audio Creators

Your listeners want more than audio. Transcripts drive search traffic, improve accessibility, and give you raw material for newsletters, show notes, and social clips. The problem is that generating and cleaning a transcript has always been tedious work — and the tools that help are usually designed for someone else's workflow.

SourceQuote is built around the recording, not the calendar invite. Upload your episode file, run transcription, and get back an accurate, speaker-attributed transcript in minutes. The waveform view makes it easy to scan through the episode visually, identify key moments, and annotate them for later use. Multiple speakers are detected and labeled automatically — give them names, assign colors, and the whole transcript becomes scannable at a glance.

Live Quotes let you pull standout moments into embeddable snippets for your website or newsletter: the guest's exact words, playable directly from the page. No clipping, no separate audio file, no hosting headache.

Projects stay organized in folders. Transcripts are shareable with editors, producers, or guests via read-only links. And because SourceQuote is free to download and run locally, there are no per-episode fees or usage caps to work around on your own hardware.

Turn every episode into more than a recording.

Privacy-Conscious Users & Organizations

Not every recording should leave the room.

Journalism sources. Legal consultations. Medical interviews. Internal strategy sessions. HR documentation. There are categories of audio that organizations have an obligation — legal, ethical, or practical — to keep under tight control. Most transcription services require uploading files to external servers. That is a non-starter in many professional contexts.

SourceQuote was designed with this constraint in mind from the beginning. The desktop application runs entirely offline on your own hardware. No account required. No network connection required. No data transmitted anywhere. Transcription is performed locally using AI models that run on your own machine.

For recordings that must be shared but not fully disclosed, SourceQuote provides content-level redaction tools that go beyond access controls. Word ranges can be muted directly in the transcript — struck from the text and excised from the audio in every shared or exported version. Paragraphs can be marked Off The Record, rendering them as a placeholder to any viewer or collaborator regardless of their permission level. Once these restrictions are in place, the original audio behind a muted or OTR section is not accessible to anyone but the editor. This is not a permission setting that can be worked around — it is enforced at the point of serving the audio itself.

For organizations that need multi-user access and collaboration, SourceQuote can be self-hosted on private infrastructure — on-premises or on a private cloud — with no dependency on the managed service at sourcequote.org. The full source code is publicly available and auditable under AGPL 3.0. There are no black boxes, no opaque data handling policies, and no vendor relationship to manage.

Security professionals, compliance teams, and cautious organizations can read every line of code that touches their data. That is not something any closed-source transcription service can offer.

Own your recordings. Own your transcripts. Own the software that processes them.

Teams & Newsrooms

Transcription is rarely a solo activity. Interviews get passed between reporters and editors. Research projects involve multiple contributors. Podcast teams include producers, writers, and hosts. The transcript that one person creates has to be reviewable, correctable, and usable by others.

SourceQuote's collaboration model is built around projects and folders, each with fine-grained access controls. Assign colleagues as viewers or editors at the project level or across entire folder hierarchies. Share a read-only presentation link with anyone who needs to read but not edit. Version history means that every change is reversible — no edit is final until you want it to be.

Access controls determine who can open a project. Content-level redaction tools determine what they can hear when they do. Editors can mute specific word ranges — removing them from the transcript text and from the audio in all shared views — or mark entire paragraphs as Off The Record, rendering them as a redaction placeholder to everyone outside the editor role. Once applied, these restrictions are enforced server-side: the audio behind muted or OTR sections is not served to viewers or shared presentation audiences regardless of their access level. The original recording stays intact for the editor; everyone else sees and hears only what the editor has approved.

For organizations running their own deployment, the admin panel provides usage analytics, user management, and subscription oversight. Scheduled jobs handle usage tracking and storage rollover automatically. Teams with demanding workloads can run parallel transcription jobs so that multiple recordings process simultaneously without blocking each other.

SourceQuote's self-hosted option means your team's data stays within your organization's infrastructure. No SaaS dependency. No per-seat licensing negotiation. A single deployment serves your entire team, at whatever scale you need.

Collaborate on recordings the way you collaborate on documents.

Technical Foundation (Non-Technical Summary)

SourceQuote is built on proven, widely-used open-source technologies. The transcription engine is OpenAI Whisper, the same model used in many commercial transcription services — except here it runs locally or on infrastructure you control. Speaker detection uses Pyannote, a research-grade speaker diarization library. The waveform interface uses WaveSurfer.js, an established open-source audio visualization library.

The backend is Python (Flask), the database is PostgreSQL (or SQLite for local use), and the frontend is plain JavaScript with no heavy framework dependencies. Cloud deployments run on Google Cloud Platform.

The result is a stack that is auditable, self-hostable, and not dependent on any proprietary AI API or vendor-locked infrastructure.

Licensing Summary

SourceQuote is licensed under AGPL 3.0 with Commons Clause.

• Free to use, modify, and self-host
• Source code is publicly available on GitLab (gitlab.com/vtleavs/sourcequote)
• Cannot be resold as a commercial product without permission
• The managed cloud service at sourcequote.org is the only paid offering; revenue covers hosting costs and supports the developer
• Donations are accepted and appreciated

Planned & Potential Features

The following features are in active planning or under consideration, drawn from the project's public issue tracker. They represent the direction SourceQuote is heading — some are near-term priorities, others are longer-horizon explorations.

Platform Expansion

• Linux and macOS desktop builds — the standalone executable currently targets Windows; Linux and macOS builds are planned and require only compilation on those platforms
• Video file support — accept video files as input, extract audio for transcription while optionally displaying the video alongside the transcript
• ASL / sign language transcription — exploratory investigation into transcription pipelines for visual rather than audio input

AI & Transcription Intelligence

• Semantic transcript search — move beyond keyword matching to meaning-based search across transcript content, allowing queries like "find where they discussed funding" rather than requiring exact terms
• Emotion detection — AI-assisted identification of emotional tone within segments, useful for tagging key moments in interviews and recordings
• LLM-assisted speaker name assignment — automatic speaker name suggestions based on conversational context, reducing the manual work of labeling unknown speakers
• Access to larger AI models — expanded access to higher-accuracy Whisper and Pyannote models for users who need the best possible transcription quality
• Contextual character segmentation — smarter segment boundary detection based on semantic and conversational context rather than silence alone
• Custom vocabulary per project — define a per-project list of proper nouns, acronyms, and domain-specific terms passed to Whisper as context, improving transcription accuracy for names, organizations, and technical language
• Global custom vocabulary — user-level vocabulary list applied automatically across all projects, with the ability to promote terms from a project vocabulary to the global list
• AI summarization and chapter markers — post-transcription LLM pass generating a plain-language summary, a list of key quotes with attribution, and timestamped chapter markers; chapter markers rendered on the waveform
• Transcription confidence visualization — per-word confidence scores displayed inline in the transcript, with low-confidence words flagged for review; includes a keyboard-navigable review mode for efficient cleanup
• Live transcription from microphone — record directly within SourceQuote and receive a live transcript as audio is captured, eliminating the record-then-upload workflow
• Transcript translation — generate a translated version of a transcript in another language, with the original and translation available as parallel views; timestamps preserved across both layers

Audio Editing & Processing

• Audio trimming — ability to trim audio directly within the workspace without external tools
• Segment and word-range audio clip export — export the audio for any segment or word selection as a standalone MP3, WAV, or FLAC file, with effects and redactions applied; useful for quote clips, highlights, and social content

Transcript Editing

• Drafter — an in-app drafting canvas for building story or article drafts from transcript excerpts; users highlight a segment or word range and add it to the Drafter panel, where excerpts can be reordered via drag-and-drop and interspersed with free-text prose; draft exportable as plain text or .docx with speaker attribution and timestamps preserved
• Insert and append to transcript — add new content at arbitrary positions in the transcript, not just edit existing segments
• Segment timing adjustment — fine-grained UI for manually adjusting segment start and end times
• Transcript highlighting and thematic tagging — apply color-coded highlight labels to arbitrary spans of transcript text for qualitative coding, theme tracking, and content categorization; tags are filterable, exportable, and summarized in a dedicated sidebar view
• SRT and VTT subtitle export — export captions in standard subtitle formats for use with video players, YouTube, Vimeo, and video editing tools; especially relevant alongside video file support

Speaker Features

• Speaker bios — attach biographical metadata to speakers within a project (role, organization, context), surfaced in exports and presentations
• Persistent speaker profiles across projects — maintain a user-level speaker database with stored voice embeddings so recurring speakers are recognized and labeled automatically across different recordings, reducing manual re-identification work

Collaboration & Integrations

• Discord integration — bot and/or webhook integration allowing transcription jobs, notifications, and transcript sharing directly through Discord servers
• Per-user change attribution — project versioning is implemented; full per-user audit trail attributing individual edits to specific collaborators is in progress
• Real-time collaborative editing — multiple editors working on the same transcript simultaneously, with live change synchronization and presence indicators, similar to Google Docs co-editing
• Slack integration — connect SourceQuote to a Slack workspace for transcription job notifications, project sharing alerts, and rich link previews when SourceQuote URLs are posted in channels
• Cloud storage import — import audio files directly from Google Drive, Dropbox, or OneDrive without downloading and re-uploading; server-side file pull avoids double-transfer through the user's machine
• Zoom and Microsoft Teams meeting import — connect a Zoom or Teams account and pull meeting recordings directly into SourceQuote for transcription, with the project name pre-populated from the meeting title

Interface & Experience

• Theme marketplace — community-contributed and curated themes beyond the built-in light/dark options, with a browsable theme selector
• Documentation versioning — version-specific documentation so users on older deployments can reference accurate documentation for their installed version
• Guide graphics and animations — illustrated onboarding and in-app guide content to make feature discovery more approachable

Developer & API

• Formalized REST API — a documented, stable public API for programmatic access to projects, transcripts, and transcription jobs
• Automated database management tools — utilities for self-hosted deployments to handle migrations, backups, and maintenance without manual intervention
• Encryption of user data at rest — end-to-end encryption of stored transcripts and audio to prevent access even at the infrastructure level

Competitive Landscape

SourceQuote operates in a field of well-established transcription and audio editing tools. The comparison below covers the five most direct competitors across the core professional use cases.

Otter.ai

What it does: AI-powered meeting and conversation transcription, built around real-time capture and team collaboration. Primarily targets business meeting use cases — Zoom calls, standups, sales calls.

How SourceQuote compares:

Otter is optimized for synchronous meeting capture — recording as the meeting happens, with minimal post-processing workflow. SourceQuote is optimized for working with finished recordings: editing, verifying, redacting, and sharing. Otter has no waveform interface, no audio editing tools, no way to mute specific words or hide segments from shared viewers, and no mechanism for embedding audio in published content. Where Otter has a clear lead is in live capture workflows: it can join a Zoom call as a bot, transcribe in real time, and surface AI-generated summaries and action items immediately after a meeting ends. SourceQuote has no live transcription mode and no mobile app. For meeting notes and quick organizational use, Otter is fast and frictionless. For journalism, research, or any workflow requiring source protection or precise audio-text work, SourceQuote is a materially different product.

Descript

What it does: Audio and video editor built around a transcript interface. The closest feature-set competitor to SourceQuote. Allows users to edit audio/video by editing the transcript text.

How SourceQuote compares:

Descript is the most capable competitor, particularly for podcast and media production. It has strong video support, screen recording, AI filler word removal, and an AI voice cloning feature (Overdub) that lets users re-record edited lines in their own synthesized voice — none of which SourceQuote offers. It also supports direct publishing to podcast platforms and has a more mature, polished UI. However, it is a closed, cloud-only product with no self-hosting option, no offline mode, and no mechanism for open inspection of how it handles data. For organizations with data sensitivity requirements — journalists with source protection concerns, researchers under IRB constraints, legal and medical contexts — the inability to self-host or run offline is a hard blocker. Descript also lacks Live Quotes, the dedicated journalism sharing workflow, and granular redaction guarantees enforced at the audio delivery layer. SourceQuote is also significantly cheaper at equivalent feature levels.

Rev.com

What it does: Transcription service offering both AI-automated and human-reviewed transcription. Produces a text file as output. No interactive workspace.

How SourceQuote compares:

Rev is a transcription delivery service, not a transcription workspace. The output is a text file; what the user does with it is outside Rev's scope. There is no waveform, no editing environment, no sharing controls, and no way to verify a quote against the source audio without switching to a separate tool. Where Rev has a genuine advantage is in accuracy for difficult audio: human transcription handles strong accents, overlapping speech, technical vocabulary, and poor recording conditions that AI models struggle with. Rev also offers translation and subtitle/caption export in broadcast-ready formats, and guaranteed turnaround windows. For users who need high-confidence accuracy on a single file with no workflow overhead, Rev is a practical choice. For users who need to work with transcripts as living documents — editing, annotating, redacting, sharing, and embedding — Rev stops where SourceQuote begins.

Trint

What it does: Cloud-based transcription platform with a transcript editor, waveform sync, and journalism-specific workflow features. One of the few competitors targeting journalists explicitly.

How SourceQuote compares:

Trint is the closest competitor in terms of target audience — it has been used in newsrooms and explicitly markets to journalists. It has a solid transcript editor with waveform sync. Beyond that baseline, Trint has features SourceQuote has not yet built: a Story Builder (Trint's name for what SourceQuote calls Drafter) that lets users drag quotes into a narrative draft, inline team commenting and review workflows, AI summarization, translation, and integrations with tools journalists already use (Adobe Premiere, Slack). These are real workflow advantages for newsrooms with existing toolchains. However, Trint lacks any mechanism for word-level audio redaction, has no Off The Record workflow, and has no equivalent to Live Quotes for embedding sourced audio in published content. It is also significantly more expensive, cloud-only, and closed-source. For a newsroom evaluating both: Trint offers a more mature product with a longer track record and a richer editorial workflow; SourceQuote offers substantially more control over source protection, substantially lower cost, and the ability to self-host or run entirely offline.

Sonix

What it does: AI transcription platform for media professionals — journalists, podcasters, filmmakers, researchers. Has an interactive transcript editor with timestamped text.

How SourceQuote compares:

Sonix has a polished transcript editor and broad language support, and its pay-as-you-go pricing can work well for infrequent users. It also has strong integrations with professional video editing tools (Adobe Premiere, Final Cut Pro, DaVinci Resolve), AI summarization, translation, and SRT/VTT export — all areas where SourceQuote has gaps. It does not have a waveform interface — navigating audio requires playback controls rather than clicking directly in a visual waveform. It has no audio redaction tools, no Live Quotes, no self-hosting option, and no offline mode. For filmmakers and video-first workflows, Sonix's NLE integrations are a practical advantage. For audio-primary work requiring source protection or visual waveform editing, SourceQuote covers ground Sonix does not.

Overall Comparison

The table below covers the full feature set across all five competitors. ✓ = available, — = not available, ~ = partial or limited, † = planned in SourceQuote roadmap.

Transcription & AI

Editing & Workspace

Audio Processing & Redaction

Sharing & Export

Collaboration & Integrations

Deployment & Privacy

SourceQuote's strongest differentiation is in three areas no competitor fully covers: source protection (word-level redaction, OTR segments, voice anonymization, offline operation), audio-anchored publishing (Live Quotes), and deployment flexibility (local, self-hosted, or managed cloud from a single open-source codebase). The most significant gaps versus leading competitors are: video support, mobile apps, real-time transcription, AI summarization, subtitle export, and the editorial workflow tools (story builder, inline comments, translation) that Descript and Trint have built for media production teams. Most of these are on the SourceQuote roadmap.

Last updated: June 2026 — planned features expanded with 14 new items