Best Free Transcription Tools, No Hidden Limits
"Free transcription" rarely means what people think. Every cloud tool has a free tier; the actual question is what's in it. This review goes through the eight free options worth knowing in 2026, with honest accounting of the per-month minutes, per-file caps, output limitations, and the sometimes-aggressive nudges to upgrade. We built one of these tools, so we'll be clear about our own free tier's limits as plainly as we describe everyone else's.
What "free" actually means in 2026
There are three kinds of "free" in this market:
- Limited free tier of a paid SaaS: most tools. Useful for trial and very-low-volume use.
- Open source you self-host: Whisper. Free forever, but you run the hardware.
- Free as in "attached to another product": built-in OS captions, YouTube auto-captions. Zero cost, narrow use case.
Each shape has its niche. Below is the honest accounting per tool.
1. MDisBetter Audio to Markdown: free account
Free tier: a free account, created in seconds; each conversion uses credits. Output: a Markdown transcript with timestamps, split into speech segments.
Strengths: a timestamped Markdown transcript is more useful for AI workflows than the plain text you get from most free tiers, and a single file can also be downloaded as .srt or .vtt subtitles. Multi-tool platform: same free-tier umbrella covers audio, PDF, URL, and 17 other converters. No watermarks, no aggressive upsell modals.
Limits to know: every conversion uses credits, so heavy users need a paid plan. No speaker identification. Web tool only, no installable app.
Best for: users feeding transcripts to ChatGPT/Claude, occasional podcast/interview transcription, anyone already using our other Markdown tools.
2. TurboScribe: free tier (the volume option)
Free tier: 3 files per day, up to 30 minutes each. That's about 90 minutes per day if you space it out, surprisingly generous.
Strengths: the 90-min/day limit is the most generous of any cloud free tier. Multiple output formats (TXT, DOCX, SRT). Polished UI. The unlimited paid plan ($10/month or so as of writing) is the best volume deal anywhere if you outgrow free.
Limits to know: 30-minute per-file ceiling means longer recordings have to be split. Plain text/SRT output, no Markdown. Some output controls gated to paid.
Best for: journalists or podcasters with multiple short files per day. Users who'll graduate to the unlimited paid plan.
3. Otter.ai: free tier (the meeting option)
Free tier: 300 minutes per month, with a 30-minute per-meeting/file cap. Real-time meeting bot included.
Strengths: 300 min/month is enough for several meetings a week. The meeting bot joins Zoom/Meet/Teams calls automatically. Speaker labels built in (check them on your own files: see how to benchmark audio to text tools). Action item extraction. Searchable archive.
Limits to know: 30-minute per-meeting cap is brutal for hour-long meetings (only the first 30 minutes are available on Basic). Free tier shows ads/upsell prompts. Plain text output.
Best for: users with regular short meetings (under 30 min). Anyone evaluating Otter before going paid.
4. Whisper local: free forever (if you self-host)
Free tier: the entire thing. Open source.
Strengths: truly unlimited. Total privacy. Every language Whisper supports is available.
Limits to know: requires Python and ideally a GPU. CPU is workable on short clips with the smaller models but painful on hour-long files. No diarization without a bolt-on (WhisperX). No web UI.
Quick start:
# Install
pip install -U openai-whisper
# Transcribe to plain text
whisper your-audio.mp3 --model large-v3 --output_format txt
# For better speed on CPU/modest hardware, use faster-whisper:
pip install faster-whisper
# In Python:
from faster_whisper import WhisperModel
model = WhisperModel("large-v3", device="cpu", compute_type="int8")
segments, _ = model.transcribe("your-audio.mp3")
for segment in segments:
print(f"[{segment.start:.2f}s] {segment.text}")Best for: developers, researchers, anyone with privacy constraints, anyone transcribing huge volumes who has a GPU.
5. Google Recorder (Pixel only): built into the phone
Free tier: unlimited, on-device, Pixel phones only.
Strengths: runs entirely offline on the phone. Real-time captioning while you record. Searchable archive. Speaker labels (limited). Free with the device.
Limits to know: Pixel phones only (Pixel 3 or newer for the modern features). English plus a handful of supported languages. Phone export workflow is fiddly if you want the file off-device.
Best for: Pixel owners doing field interviews, meeting notes, voice memos.
6. macOS Live Captions: built into the OS
Free tier: unlimited, on-device, macOS Sonoma+ and recent iOS.
Strengths: truly free, runs offline, works on any audio playing on the device (FaceTime call, YouTube video, podcast app). Real-time. Privacy-friendly.
Limits to know: caption-style output only; short rolling lines, not a saved transcript file by default. Capturing the full transcript requires manual scrollback or a screen-recording trick. Limited languages. Apple Silicon recommended.
Best for: live captioning of calls or media. Less useful for archival transcription.
7. YouTube auto-captions: the free trick
Free tier: unlimited, requires uploading audio as an unlisted YouTube video.
How it works: upload your audio (as a video, pair it with a static image) as an unlisted YouTube video. Wait for YouTube to auto-generate captions (usually within an hour). Open the video, click the three-dot menu under the player, choose "Show transcript." Copy the transcript text.
Strengths: truly free. Surprisingly good accuracy on clear English audio (Google's models are strong). Handles long files (multi-hour videos work fine).
Limits to know: upload + processing time (often 30-60 minutes). Requires a Google account. Captions are time-stamped by line, not paragraph-formatted. Plain text, no speaker labels, no Markdown structure. Privacy implication: your audio is on YouTube's servers (unlisted but accessible to anyone with the link).
Best for: users with no budget who don't mind the upload latency and the privacy tradeoff.
8. VOMO: free tier
Free tier: limited monthly minutes, mobile-app focused.
Strengths: structured Markdown output (MDisBetter is one of the few others that returns Markdown). Mobile-first capture.
Limits to know: smaller free quota than TurboScribe or Otter. Web experience secondary to mobile.
Best for: mobile-heavy capture, users who want Markdown output on the go.
Honorable mention: Trint trial
Trint offers a 7-day free trial with full features, no free-forever tier. Useful for one-off short projects (a single conference's worth of audio, for example) but not a sustainable free option.
Free-tier comparison table
| Tool | Monthly limit | Per-file cap | Output | Diarization | Best for |
|---|---|---|---|---|---|
| MDisBetter | Free credits (then paid) | Generous | Markdown (timestamped), SRT, VTT | No | AI workflows |
| TurboScribe | ~90 min/day | 30 min | TXT, SRT, DOCX | Yes | Daily small files |
| Otter | 300 min | 30 min | Plain text | Yes | Short meetings |
| Whisper local | Unlimited | None | Any (TXT, JSON, SRT, VTT) | With WhisperX | Volume + privacy |
| Google Recorder | Unlimited (Pixel) | None | App-only | Limited | Pixel users |
| macOS Live Captions | Unlimited | N/A | Live captions | No | Live calls/media |
| YouTube auto-captions | Unlimited | None | Plain text | No | No-budget option |
| VOMO | Limited | Limited | Markdown | Yes | Mobile capture |
Recommendations by use case
Student transcribing a lecture: Whisper local if you have a laptop with reasonable specs; MDisBetter free tier if you want a timestamped Markdown transcript for studying with ChatGPT; YouTube auto-captions trick for long lectures with no budget.
Journalist with a one-off interview: MDisBetter or Trint trial. If accuracy is critical for a published piece, the AI tier of HappyScribe (paid) or human transcription is worth budgeting for.
Podcaster with daily recordings: TurboScribe free (3 files of 30 min/day) is the most generous. Graduate to TurboScribe unlimited (~$10/month) if volume scales.
Sales team with weekly meetings: Otter free tier (300 min). Watch the 30-min per-meeting cap; clip long calls or upgrade.
Developer batch-processing many files: Whisper local, run via faster-whisper for speed. Free at any volume.
Privacy-critical work: Whisper local, no cloud option is acceptable. Period.
What you give up at the free tier
Across the cloud tools, the most common things gated to paid are: longer per-file caps, more monthly minutes, additional languages, downloadable formats beyond plain text, advanced editor features, team collaboration, real-time captioning, and (sometimes) accuracy; some vendors route free-tier audio through faster, slightly-less-accurate models.
The format limitation often hurts the most. Most free cloud tiers give you plain text or basic SRT. If you're feeding the transcript to an LLM, a Markdown transcript with timestamps is meaningfully better: the model can point to the moment each thing was said, and you can add section headings yourself if you want to navigate by topic. We unpack the difference in speech to text vs audio to Markdown.
What about other free Markdown utilities?
If you're working with audio you're often working with PDFs and web articles too. The same free-platform logic applies: see the parallel best free PDF to Markdown converters review. Routing everything through Markdown means the same downstream tools (chunker, embedder, retrieval) work for the whole corpus.
The honest summary
If you have any technical comfort, Whisper local is the unbeatable free option for high volume. If you don't, TurboScribe free has the most generous cloud quota for daily use. If your endpoint is an AI tool, MDisBetter gives you timestamped Markdown on a free account. If meetings are your job, Otter free covers most cases until you hit the cap. The mistake to avoid is paying for a tool whose free tier you haven't tested first: every tool above gives you enough free runway to evaluate properly before committing.
Check Otter Basic limits: 300 monthly minutes, 30 minutes per transcript and three lifetime file imports.