Home / Guides / How to transcribe audio to text
How to transcribe audio to text
There are four practical ways to turn a recording into text. They differ in cost, speed, accuracy and in who gets a copy of your audio.
The fastest free way to transcribe audio to text is an on-device tool: open one in your browser, choose the recording, select the spoken language and let a speech model convert it on your computer. Then read through the result while listening and correct names, numbers and unclear passages. Typing by hand, phone dictation and paid cloud services are the alternatives.
The four methods at a glance
| Method | Cost | Your effort | Where the audio goes |
|---|---|---|---|
| Type it yourself | Free | Several hours per hour of audio | Stays with you |
| Dictation built into your phone or computer | Free | Low for short clips; awkward for files | Depends on the device and settings |
| Cloud transcription service | Free tier with limits, then a subscription or per-minute fee | Low | Uploaded to the provider |
| On-device tool such as QuietSteno | Free | Low | Stays on your device |
Method 1: type it yourself
Play the recording and type what you hear. It needs no tools and gives you full control, which is why it is still used for short clips and for difficult audio. It is slow: people speak several times faster than they type, and you have to stop and rewind constantly, so an hour of audio commonly takes around four hours.
If you do type, slow the playback to about 70 percent and use a player with keyboard shortcuts for pause and skip-back, so your hands stay on the keys.
Method 2: built-in dictation and recorder apps
Phones and computers include speech recognition: voice typing on the keyboard, and recorder apps on some recent phones that transcribe as they record. These are good for capturing new speech. They are less suited to an existing file, because many dictation features only listen to the microphone and offer few ways to export timestamps or subtitles.
Method 3: cloud transcription services
You upload the file, the provider’s servers transcribe it, and you read or download the result. Large server models are the most accurate option available, and many services add speaker labels, summaries and team features.
The costs are a monthly or per-minute price beyond a limited free tier, the time to upload, and the fact that your recording is now stored on someone else’s systems. For confidential audio, read the provider’s retention and training terms first. See is online transcription private?
Method 4: on-device transcription in the browser
Open-source speech models such as Whisper are now small enough to run inside a web browser. The page downloads the model once, then processes recordings on your own processor or graphics chip. There is nothing to install, no account, and no per-minute cost, because no server does the work.
With QuietSteno the steps are:
- Choose the recordingDrag an audio or video file onto the pad. MP3, WAV, M4A and MP4 are all fine.
- Select the spoken language and accuracyBalanced suits most recordings. Use Most accurate for accents, noise or languages other than English.
- Wait for the transcriptThe first run downloads the model. Lines then appear in order, each with a timestamp.
- Proofread while listeningPress play, follow the highlighted line and fix what was misheard.
- ExportCopy the text, or download TXT, SRT or VTT.
Whatever the method: proofread
No automatic transcript is perfect. Errors concentrate in a few places, so check these first:
- Names of people, companies, places and products.
- Numbers, dates, amounts and units.
- Negatives: “can” and “can’t” sound alike and reverse the meaning.
- Specialist terms the model may replace with a common word that sounds similar.
- Passages with crosstalk or noise, where words may be missing.
Recording so that it transcribes well
The biggest gains come before transcription. Put the microphone close to the speaker, record in a quiet room with soft furnishings, avoid clipping, and ask people not to talk over each other. A clear recording on the Balanced setting beats a poor one on any setting.
Questions and answers
What is the easiest free way to transcribe audio to text?
Use an on-device browser tool. Choose the file, select the language and wait for the text. There is no sign-up, no minute limit and no upload.
How long does it take to transcribe one hour of audio?
By hand, around four hours for most people. Automatically, it depends on the tool and hardware: cloud services and computers with a capable graphics chip are usually faster than real time, while older devices take longer.
Can I transcribe audio to text without uploading it?
Yes. On-device tools run the speech model on your own computer, so the recording is never transmitted. QuietSteno works this way.
How accurate is automatic transcription?
For clear speech from one person, most words are right and proofreading is quick. Accuracy falls with background noise, overlapping voices, strong accents and specialist vocabulary.
Related
- TranscribeAny recording to an editable transcript.
- Audio to textMP3, WAV, M4A, OGG or FLAC to text.
- Choosing an accuracy settingWhat Fast, Balanced and Most accurate mean.
- Is online transcription private?Where your recording goes, and how to check.
Last updated 13 November 2026.