HablakiHow to transcribe audio to text
- Choose your file. Drag it into the box or click to browse for it. Recordings, meetings, classes, interviews, podcasts, videos and voice notes all work.
- Select the language and the mode. Hummingbird is the fastest, Lynx balances speed and accuracy, and Owl is the most accurate.
- Press “Transcribe”. You'll see the text appear as it progresses. When it's done, correct it if needed and download it in the format you need.
Fast in the cloud, or private on your device
You can transcribe in two ways. In the cloud, the audio travels encrypted to be transcribed in seconds and is deleted when it's done: we store neither the audio nor the text. On your device, speech recognition runs inside your browser and the file never leaves your computer. Either way, no sign-up and no email required.
| Criterion | This transcriber | Other common services |
|---|---|---|
| Where the audio is processed | You choose: in the cloud or on your device | On their servers |
| Your files | Not stored | Usually kept in your account |
| Account | Not needed | Usually required |
| Free usage limit | 5 transcriptions a day, up to 1 hour each | Usually just a few files or minutes a day |
| Speed | In the cloud, seconds; on your device, depends on your computer | Depends on the queue and your connection |
Transcription modes
| Mode | Best for | First-time download |
|---|---|---|
| Hummingbird | Voice notes and short, good-quality audio, or computers without a graphics card. | About 150–200 MB |
| Lynx | The recommended option: meetings, classes and interviews. | About 300 MB |
| Owl | Difficult audio, strong accents or technical vocabulary. Needs a compatible graphics card. | About 560 MB |
Input and output formats
Accepts MP3, MP4, M4A, MOV, AAC, WAV, OGG, OPUS, MPEG, WEBM, FLAC, WMA and WMV, plus WhatsApp and Telegram voice notes. Videos work too: only their audio is used.
Download the result as text (TXT), a Word document (DOCX), PDF or subtitles (SRT and VTT) for YouTube, Premiere Pro or DaVinci Resolve. You can also copy it or summarize it with ChatGPT.
What it is for
- Meetings and calls: turn your Zoom, Meet or Teams recording into minutes you can review and share.
- Classes and lectures: get your notes in writing and find what the professor said in seconds.
- Interviews and journalism: transcribe with separate speakers and quote without listening to everything again.
- Podcasts and videos: create subtitles, descriptions and articles from what you recorded.
- Voice notes: read that long WhatsApp audio when you can't listen to it.
Tips for a more accurate transcription
- Pick the right audio language: it's what affects the result the most.
- If the audio is noisy or the speaker has a strong accent, use Owl.
- For meetings with several people, turn on “Identify speakers”.
- Use a computer with Chrome or Edge: it's much faster than on a phone.
What if, instead of recording and transcribing, you dictated directly?
With Hablaki, you press a key, speak, and the text appears wherever your cursor is: in your email, in Word, in WhatsApp Web or in any app. Your voice never leaves your device. Try it free for 14 days, no card required.
Download free · 14-day trialSee how it worksFrequently asked questions
Is it really free?
Yes. No sign-up and no watermark: up to 5 transcriptions a day, up to 1 hour each. If you need more, the Hablaki app has no limits.
Is my file stored?
No. In the cloud, the audio travels encrypted, is transcribed and is deleted when it's done: we store neither the audio nor the text. If you'd rather it not leave your computer, choose “On your device” and it will be transcribed inside your browser.
How long can the audio be?
Up to 1 hour per file, with files up to 2 GB. For longer recordings, use the Hablaki app, which has no length limit.
What formats does it accept?
MP3, MP4, M4A, MOV, AAC, WAV, OGG, OPUS, MPEG, WEBM, FLAC, WMA and WMV, plus WhatsApp voice notes.
Does anything get downloaded to my computer?
In the cloud, nothing. If you choose “On your device”, the speech recognition model is downloaded once: an artificial intelligence file that lets your browser understand the audio. No program is installed.
How much space does it take?
It depends on the mode: Hummingbird, between 150 and 210 MB; Lynx, about 300 MB; and Owl, about 565 MB. It's stored within the browser's storage, not in your folders.
Do I have to download it every time I visit?
No. It stays saved in your browser and starts instantly the next times, even if you reload the page or come back another day. It's only downloaded again if you clear your browsing data, use incognito mode, or switch browsers or devices.
How do I free up that space?
Press “Free up space”, below the transcribe button. It's also deleted if you clear this site's data in your browser settings.
Is there a lighter option?
Yes: Hummingbird mode is the lightest. It's ideal for voice notes and clear audio; for noisy audio or several speakers, Lynx gives better results.
Does it work without internet?
You need a connection to open the page. In “On your device” mode, once the model is downloaded, transcription happens on your computer.
Where are my transcriptions saved?
Only in your browser, on this device. No one else can see them, not even us. If you clear your browsing data they're lost, so download the ones you want to keep.
Can I transcribe from my phone?
It works on some recent phones, but it's much slower. For files longer than a few minutes, use a computer with Chrome or Edge.
How accurate is it?
It uses one of the most accurate speech recognition models available. With clear audio it gets almost every word right; with noise, strong accents or several people talking at once, it makes more mistakes. Always double-check anything important.