Free transcription software is not one category.

A local desktop app can transcribe without a monthly quota. A browser tool is faster for one audio file. A meeting assistant can join scheduled calls automatically, but usually stores the result in a cloud workspace. Those are different workflows, even when every product calls itself free.

The short version:

• Choose Anarlog for live, bot-free meeting transcription with local data ownership and no transcription-minute cap.
• Choose MacWhisper for private audio or video file transcription on a Mac.
• Choose Riverside for an occasional browser upload without creating an account.
• Choose Transcript LOL when the source is a URL or cloud file and two short transcripts per day are enough.
• Choose Otter for automatic calendar meetings and a searchable cloud meeting workspace.
• Choose Notta for multilingual cloud transcription, while treating its three-minute free-session limit as an evaluation tier.
• Choose oTranscribe for a permanently free manual program with zero cloud upload, on any desktop OS.
• Choose Whisper or whisper.cpp for a free automatic engine you run yourself, if you are comfortable with a command line.

Free transcription software compared

Start with what you need to capture, where you are willing to process it, and whether the free limit survives normal use.

How we evaluated these tools

We used five practical filters:

1. The free access is ongoing. A permanent free plan, open-source app, or unlimited free utility qualifies. A short trial does not.
2. The limit is explicit. Minutes, file imports, meeting duration, and export restrictions matter more than a large "Free" label.
3. The capture method fits the job. Live meetings, uploaded recordings, browser URLs, and manual transcription are different jobs.
4. The data path is understandable. Local tools keep processing on your computer by default. Cloud tools require you to evaluate the vendor's current privacy and retention terms.
5. The upgrade trigger is honest. We call out the point where a free plan stops being useful instead of hiding it below a feature list.

We checked each vendor's current product, pricing, and privacy pages for this update. Plans still change often, so follow the source links before committing a workflow.

1. Anarlog: best for free live meeting transcription with local control

Anarlog is an open-source AI meeting notetaker for macOS, Windows, and Linux. It captures microphone and system audio from your computer, so no meeting bot needs to join Zoom, Microsoft Teams, Google Meet, or an in-person conversation.

Anarlog keeps a structured meeting summary beside the local meeting history.

The canonical meeting record lives in local SQLite. Notes, transcripts, and meeting metadata are not dependent on a vendor-hosted workspace, and Markdown is available when you want a portable export.

What the free plan includes

• Unlimited local transcription with no monthly minute quota
• Bot-free capture of microphone and system audio
• Local storage for meetings, notes, and transcripts
• Bring-your-own-key support for AI providers
• Local AI through tools such as Ollama or LM Studio
• Search, templates, imports, and portable exports
• An MIT-licensed codebase that can be audited or forked

The free path is most useful when you already have an API key or are comfortable running a local model for summaries. Local transcription itself does not require a paid Anarlog subscription.

When you would pay

Anarlog Pro is $15 per month or $150 per year when you want hosted transcription and AI, encrypted CloudSync, sharing, and calendar-connected services without configuring your own providers.

2. MacWhisper: best for private audio and video files on Mac

MacWhisper is a native Mac app built around local speech-to-text models. Drop in a recording, record from the microphone, or use its basic dictation mode, and the transcription runs on your Mac by default. Our Mac transcription software comparison covers the broader choice between file, dictation, and live-meeting workflows.

MacWhisper's current product page shows a local transcript beside its optional AI chat panel. Source: MacWhisper.

Its free version includes smaller Whisper models, more than 100 languages, and TXT, SRT, and VTT export. Because the audio stays on the device when you use local models, it is a strong fit for interviews, lectures, and recordings you do not want to upload to a browser service. MacWhisper's privacy documentation explains the exceptions: cloud transcription, translation, or AI providers receive data when you explicitly use those services.

What the free version includes

• Unlimited local file transcription
• Tiny, Base, and Small Whisper models
• Microphone recording and basic dictation
• TXT, SRT, and VTT export
• Local processing on macOS

When you would pay

MacWhisper Pro is a one-time purchase through the official site. It adds larger and faster models, batch transcription, advanced exports, automatic speaker recognition, meeting recording, real-time captions, and more AI-provider integrations. The Gumroad and Mac App Store editions have different licenses and pricing, so buy through MacWhisper's official site rather than a lookalike domain.

3. Riverside: best for a one-off transcript without signup

Riverside's standalone transcriber accepts audio or video in a browser and does not require an account. Riverside says the utility is free to use without a transcription quota, supports more than 100 languages, and exports TXT or SRT.

Riverside's first-party transcription page shows a speaker-labeled transcript generated from recorded audio. Source: Riverside.

That makes it the lowest-friction option in this list for a single non-sensitive file. Upload, wait for the transcript, download it, and leave.

The trade-off is context. This is a cloud upload, not an on-device workflow. The free standalone tool also assigns all text in a multi-speaker file to one speaker. Riverside's paid recording platform adds speaker detection, editing, captions, and the rest of its creator workflow.

Riverside's privacy policy treats recordings and transcripts as Content Data and describes how that data may be used to provide and improve the service. Read the current policy before uploading confidential interviews or meetings.

4. Transcript LOL: best for short files and media URLs

Transcript LOL is a cloud transcription tool designed around flexible input. It accepts files, cloud drives, and links from media platforms. Its free tier allows two transcripts per day, each up to 20 minutes, with TXT, DOCX, PDF, SRT, and VTT exports.

Transcript LOL's product example shows automatic speaker detection inside the transcript. Source: Transcript LOL.

That daily reset can be more useful than a one-time trial. It works well for short clips, voice notes, and individual podcast or video segments. It is less useful for full-length interviews because the 20-minute limit applies to every free upload.

What the free plan includes

• Two transcripts per day
• Up to 20 minutes per transcript
• One upload processed at a time
• File, cloud-drive, and supported URL inputs
• Speaker detection
• TXT, DOCX, PDF, SRT, and VTT export

Summaries, custom prompts, chat, longer files, and priority processing require the paid plan. Transcript LOL says in its privacy policy that it does not use customer content to train AI models, but the files are still processed through a cloud service.

5. Otter: best for automatic meeting capture on a free plan

Otter is no longer positioned as a simple transcriber. It describes itself as a conversational knowledge engine that turns meetings into a searchable workspace and pushes the output into other tools.

Otter's official Notetaker overview shows the calendar, visible guest, and transcript workflow. Source: Otter.

Its Basic plan includes 300 transcription minutes per month, with a 30-minute cap per conversation. Otter Notetaker can join Zoom, Google Meet, and Microsoft Teams as a participant. Otter also offers bot-free recording through desktop, mobile, and browser capture paths, so the bot is an option rather than the only workflow.

What the free plan includes

• 300 transcription minutes per month
• Unlimited meetings, up to 30 minutes per conversation
• Three audio or video file imports for the lifetime of the account
• The 25 most recent conversations in history
• Speaker identification, editable text, playback, and basic AI Chat
• TXT and MP3 export

The free plan is generous for short recurring calls but poor for a backlog of recordings because file imports are limited to three for the life of the account. PDF, DOCX, SRT, longer meetings, more imports, and deeper workflow features require a paid plan. Verify the current numbers on Otter's pricing page.

Otter is a cloud workspace. Its privacy policy describes using de-identified audio recordings and transcriptions to improve and train its proprietary AI technology. If that boundary does not fit your meeting, use a local workflow instead.

6. Notta: best for testing multilingual cloud transcription

Notta combines real-time transcription, meeting bots, a desktop app, mobile capture, file uploads, translation, and post-meeting AI deliverables. Its clearest advantage is language breadth: Notta currently advertises transcription in 58 languages and supports several bilingual and translation workflows on paid plans.

Notta's first-party product image shows the same meeting record across desktop and mobile. Source: Notta.

The free tier looks larger in minutes than it feels in use. Notta's public pages show different monthly allowances by region, including 120 and 200 minutes, but each free conversation stops at three minutes. That is enough to test quality on your own accent and audio setup. It is not enough for a normal meeting or interview.

What the free plan includes

• A monthly transcription allowance, with regional variation
• Three minutes per conversation
• Up to 50 file uploads per month
• A limited number of AI summaries
• Real-time transcription, mobile apps, and speaker identification

Paid plans raise the recording duration, add transcript export and translation capabilities, and expand collaboration and integrations. Check Notta's pricing page because the regional free allowance and paid pricing can differ.

Completely free manual and open-source programs

The six tools above are all automatic: you give them audio, and they produce a transcript with no typing. That is not the only kind of genuinely free program. If you want something you can install once, keep forever, and never pay for even at scale, two options are worth knowing about, with a real trade-off attached to each.

oTranscribe: manual transcription with zero cloud upload

oTranscribe is a free, open-source browser app built by journalist Elliot Bentley as a MuckRock Foundation project. It does not transcribe automatically. It puts an audio or video player and a text editor in the same window so you type your own transcript without switching applications. For someone on Windows or Linux searching for "a free program to transcribe audio to text", where MacWhisper is not an option, this is the direct answer.

oTranscribe's own app screen shows the file/YouTube import step and the manual transcript editor. Source: oTranscribe.

What it includes

• No signup, no install beyond a browser, and no daily or monthly limit
• The audio file and transcript never leave your computer
• Keyboard shortcuts to pause, rewind, and insert a timestamp without leaving the keyboard
• Autosave to browser storage as you type
• Export to Markdown, plain text, or Google Docs
• MIT-licensed and open source

The trade-off is in the name: there is no speech-to-text model here, so you are typing the transcript yourself while the player does the pausing and rewinding for you. Use it when privacy matters more than speed, or when one file needs doing today and an account is not worth creating.

Whisper and whisper.cpp: free automatic transcription for developers

OpenAI's Whisper is an open-source speech-recognition model, and whisper.cpp is a widely used C/C++ port that runs it efficiently on ordinary laptop hardware, including Apple Silicon and consumer Windows and Linux machines. Both are free to download and run locally, with no per-minute cost and no account.

Unlike oTranscribe, this is automatic: point it at an audio file from the command line and it produces a transcript entirely on your machine. The catch is packaging, not capability. Running the raw model means installing dependencies, picking a model size for your hardware, and working from a terminal instead of a graphical app. MacWhisper exists precisely to remove that friction: it wraps the same class of local model in a normal interface, at the cost of being Mac-only and partly paid instead of free and cross-platform.

Rev, Sonix, Trint, and Happy Scribe are a different case. Each offers a trial or a small starter allowance, and the automatic transcription becomes paid the moment that credit runs out, so none of them belong on a genuinely free list either.

Which free transcription tool should you choose?

Use the input and privacy boundary to decide:

• Live meetings, local record: Anarlog
• Private files on Mac: MacWhisper
• One quick browser upload: Riverside
• Short online media or cloud files: Transcript LOL
• Automatic scheduled meetings: Otter
• Multilingual cloud evaluation: Notta
• Permanently free manual program, any OS: oTranscribe
• Permanently free automatic program you run yourself: Whisper or whisper.cpp

For sensitive material, do not stop at a product's marketing page. Confirm consent requirements, the exact capture path, enabled AI providers, retention settings, and the current privacy agreement for every service that receives the audio.

Frequently asked questions

Is there completely free transcription software?

Yes. Anarlog provides unlimited local meeting transcription, MacWhisper provides unlimited local file transcription on Mac, Riverside offers a free standalone browser transcriber, and oTranscribe and whisper.cpp are free and open source on any desktop OS. The catch differs: local apps need your computer to run the model, a browser tool requires uploading the file to a cloud service, and oTranscribe specifically is manual rather than automatic.

What is the best free transcribing software for Mac?

Use MacWhisper for existing audio and video files. Use Anarlog for live meetings, system audio, notes, and a searchable local meeting history. Both offer ongoing local transcription without a monthly minute bucket, but they solve different jobs.

Can I transcribe audio for free without uploading it?

Yes. Use an on-device tool such as MacWhisper or whisper.cpp for files, oTranscribe for manual transcription, or Anarlog for live meetings and imported recordings. Keep the speech-to-text model and any summary model local if the audio must not leave your computer.

Which free transcription software works for meetings?

Anarlog captures meetings from your desktop without a bot. Otter can automatically join scheduled meetings or record through its desktop and browser tools. Notta also supports meeting bots and desktop capture, but its free three-minute conversation limit makes it better for testing than routine meetings.

Are free transcription tools accurate?

Accuracy depends more on the recording than the price label. Clear audio, separate microphones, limited crosstalk, and the right language model matter. Treat vendor percentage claims as marketing unless they publish a reproducible benchmark. Test the tool with your own speakers, accents, terminology, and worst normal audio before building a workflow around it.

What is a free program to transcribe audio to text?

It depends on whether you need automatic speech-to-text or are willing to type it yourself. For a fully automatic, permanently free program, use Anarlog for live meetings or MacWhisper for existing files on a Mac; on Windows or Linux, whisper.cpp runs the same class of open-source model from the command line. For a completely free manual option on any desktop OS, use oTranscribe: nothing uploads, there is no account, and you control playback while you type.