AI transcription — free tool
Free AI transcription: upload audio or video and get a TXT or SRT transcript using self‑hosted Whisper‑based models — no credit card required.
Built for teams that want transcripts to turn into reusable, searchable assets.
AI transcription — free tool
Free AI transcription means you can upload audio or video and get a usable transcript in minutes, without paying upfront. With Wisprs, the free flow lets you upload common formats, choose a speed or quality setting, and click “Start transcribing” to generate TXT or SRT output. It runs on a self‑hosted, Whisper‑based pipeline (faster‑whisper small or large‑v3) and supports language auto‑detection. You do not need a credit card to try it. Limits are real: no speaker diarization on free, exports may include a watermark, and accuracy varies with audio quality and language.
Start now: Start transcribing
How to use the free flow right now
The free workflow is designed for quick, low-friction transcripts. You upload a file, confirm your settings, and start the job. Processing runs asynchronously in a queue dedicated to the free tier, so short files complete quickly while longer ones may take more time.
After upload, you can choose how the system balances speed and quality. “Speed” favors faster turnaround using lighter settings, while “Best quality” uses a larger model configuration for clearer results on difficult audio. “Auto” lets the system pick a reasonable default.
Once the transcript is ready, you can edit it in the dashboard and export as TXT or SRT. If you spot mistakes, fix them directly and re‑export without rerunning the job.
- Upload your audio or video file, then confirm to proceed
- Choose Speed, Auto, or Best quality for the free self‑hosted path
- Click “Start transcribing” and wait for completion in the queue
- Open the transcript, edit any lines, and export as TXT or SRT
This flow is enough for quick notes, captions, or a first pass you plan to refine.
Supported inputs and outputs
Wisprs accepts a wide range of common audio and video formats, so you can bring files from phones, recorders, or editing tools without conversion. The system performs language auto‑detection across 100+ languages, which helps when you are unsure about locale settings or are working with mixed content.
On the free plan, export options are intentionally simple. You get plain text for reading and SRT for subtitles. These formats cover most immediate needs, such as sharing a transcript or adding captions to a video.
- Supported inputs: AAC, FLAC, M4A, MP3, MP4, MPEG, MPGA, OGG, WAV, WEBM
- Language auto‑detection for 100+ languages
- Free exports: TXT and SRT (subtitle-ready)
If you need more advanced formats or structured outputs, those are part of paid plans and are covered later on this page.
What the free workflow includes and doesn’t
The free tier is built to be useful on its own while staying transparent about tradeoffs. It uses a self‑hosted bridge with faster‑whisper models and processes jobs in a queue labeled `transcription-free-self-hosted`. This keeps the experience accessible without requiring payment, but it also means performance can vary during busy periods.
Accuracy is generally strong on clear audio, especially with minimal background noise and a single speaker. It can drop with cross‑talk, heavy accents, or poor recording conditions. That is normal for speech recognition systems and not unique to this tool.
Here is what you should expect on free:
- Engine: self‑hosted faster‑whisper (small or large‑v3), queued processing
- Controls: Speed vs Quality (Auto, Speed, Best quality)
- Editing: in-dashboard transcript editing and re‑export
- Exports: TXT and SRT; exports may include a watermark
- Not included: speaker diarization (no automatic speaker labels)
Real‑time transcription is available in the product via a WebSocket endpoint, but most free users will rely on file uploads for predictable results. For continuous or live workflows, paid plans provide a more reliable setup and routing.
Where free workflows usually break
Free transcription is best for short, straightforward files. As your needs grow, you will likely run into a few common limits. These are not hidden constraints; they reflect the difference between a quick utility and a production workflow.
Long recordings can take longer to process in a shared queue, and large files are more sensitive to network interruptions. Multi‑speaker content is harder to read without diarization, since all text appears in a single stream. If you need clean captions with timing control or structured data for editing, TXT and SRT may feel basic.
- Long files may process slowly or require patience during peak times
- Multiple speakers are not separated on free (no diarization)
- Advanced export formats (like VTT, DOCX, JSON) are not included
If any of these issues block your workflow, it is a sign you are moving beyond a quick transcript and into repeatable production.
When to upgrade to a richer workflow
Upgrading makes sense when you need consistency, speed at scale, or features that save editing time. Paid plans route transcription through premium engines (such as ElevenLabs Scribe), which can improve handling of longer files and support features like speaker identification.
You also create batch processing, broader export formats, and AI features that turn transcripts into usable assets. Summaries, chapters, and action items help you move from raw text to something you can publish or share quickly. For teams, collaboration and higher limits reduce friction across projects.
- Speaker identification (diarization) for interviews and meetings
- Batch uploads and parallel processing for multiple files
- Additional exports (for example VTT, DOCX, JSON)
- AI summaries, chapters, and action items
- More predictable performance for longer recordings
If you are testing the tool, start free and upgrade only when your use case demands it. See details on pricing and a full feature breakdown on features.
Real examples: how people use the free tool
A podcaster might drop a 20‑minute episode clip to generate SRT captions for social video. They choose “Best quality,” wait for the job to finish, then export SRT and make light edits for timing. It is fast enough for weekly posts without a paid plan.
A student might upload short lecture segments to create notes. They export TXT, clean up terminology in the editor, and keep a searchable archive for exams. For longer lectures or multiple files at once, they may consider upgrading.
An interviewer might transcribe a single recorded conversation to pull quotes. Even without speaker labels, the transcript is good enough to identify key passages. If they need clear speaker separation across many interviews, diarization becomes the reason to upgrade.
Accuracy, processing, and what affects results
Accuracy depends on the audio more than anything else. Clean recordings with a single speaker, minimal background noise, and consistent volume produce the best transcripts. Overlapping speech, strong accents, or low bit‑rate files can reduce clarity.
The free tier uses faster‑whisper models in a self‑hosted environment. “Best quality” typically yields better results on difficult audio, but it may take longer than “Speed.” Language auto‑detection helps avoid incorrect locale settings, though you can still see mixed results in multilingual clips.
If you need consistently higher performance on long files or multi‑speaker content, paid routing uses premium engines and adds diarization. That combination reduces editing time, which is often the real cost in transcription workflows.
For a practical walkthrough of improving results, see this guide: Check transcript quality.
FAQ
Is this really free to use? Yes. You can upload files and generate transcripts without a credit card. The free plan includes TXT and SRT exports, with some limits such as no diarization and possible watermarks.
What file types can I upload? Common audio and video formats are supported, including MP3, WAV, M4A, MP4, OGG, and WEBM, among others. You generally do not need to convert files before uploading.
How accurate is the transcription? Accuracy is strong on clear audio and can vary with noise, accents, and overlapping speech. Using “Best quality” can improve results on harder recordings, though it may take longer.
Can I identify different speakers? Not on the free tier. Speaker identification (diarization) is available on paid plans and is useful for interviews, meetings, and podcasts.
What formats can I export for free? You can export TXT for reading and SRT for subtitles. Additional formats are available on paid plans.
Do I need to install anything? No. The free tool runs in your browser. Upload your file, start the job, and download the result.
Is real-time transcription available? The product supports real-time transcription via a WebSocket endpoint. Most free users rely on file uploads for predictable results, while real-time and higher-throughput use cases are better served on paid plans.
Can I edit the transcript? Yes. You can edit the transcript in the dashboard and re‑export without rerunning the job.
Does it support multiple languages? Yes. The system includes language auto‑detection across 100+ languages, which helps when working with varied content.
What happens if my file is long? Longer files may take more time in the free queue. If you regularly process long recordings or many files, a paid plan will be more reliable.
Start transcribing for free
You can get a usable transcript in a few clicks, then decide if you need more. The free path covers quick captions, notes, and one‑off interviews, while paid plans create scale and advanced features when you are ready.
Primary: Start transcribing Explore: View pricing · See all features · Read the guide