How to Transcribe Audio to Text

You can transcribe an hour of audio for $0, with no upload and no account. It's easy to miss, because the search results for this question are wall-to-wall upload sites with free tiers that cap out fast. So we lined up seven methods, checked every free limit on the vendor's own page, and sorted them by what you're actually starting with.

The Quick Answer

Method 1. Audacity and Whisper, Free and Offline

This one's the quiet winner. Audacity, the free audio editor, can run OpenAI's Whisper speech model through a free plugin. Per Audacity's transcription page, it works on Windows, Mac, and Linux, runs entirely on your machine, and has no minute caps, subscriptions, or watermarks.

1. Install Audacity, then install the free OpenVINO AI plugin from Audacity's official plugins page.
2. Open Audacity and import your recording.
3. Select the track, or just the part you want transcribed.
4. Open the Analyze menu and choose OpenVINO Whisper Transcription.
5. Pick a model and a language, then click Apply.
6. Export the result as plain text, SRT, or VTT.

The transcript lands in a new label track with timestamps. The trade is speed. Bigger Whisper models are more accurate and slower, and Audacity notes they run slower on a CPU. There aren't any speaker labels either. For a long lecture or a private interview, none of that matters much next to free and offline.

Method 2. Microsoft Word's Transcribe Button

If you're paying for Microsoft 365, you've already got a decent transcriber. Microsoft's support page says subscribers can transcribe up to 300 minutes of uploaded audio a month, from .wav, .mp4, .m4a, or .mp3 files, in Word for the web and Word for Microsoft 365 on Windows in business accounts.

1. Open a document in Word for the web.
2. On the Home tab, open the Dictate dropdown and choose Transcribe.
3. In the Transcribe pane, select Upload audio and pick your file.
4. Wait for it to finish, then rename the speakers if you like.
5. Choose Add all to document to drop the full transcript into the page.

It'll separate voices and label them Speaker 1, Speaker 2, and so on, and you can rename every instance at once. The catch is the subscription. If you don't already have one, 300 minutes isn't worth signing up for when Method 1 is free.

Method 3. Google Docs Voice Typing, for Live Speech

Google Docs can't take an audio file. What it does is type what you say into the microphone, in real time. Google's help page says it works in current versions of Chrome, Edge, and Safari, and supports a long list of languages. You can say punctuation like "period" and "new paragraph", though Google notes punctuation varies by language and editing commands work in English only.

1. Open a document in Google Docs.
2. Click Tools, then Voice typing.
3. Click the microphone when you're ready.
4. Speak clearly at a normal pace, saying punctuation out loud.
5. Click the microphone again to stop.

Some people play a recording through their speakers into the mic to fake a file upload. It'll work for a clean single voice. Expect plenty of fixes on anything else.

If what you really want is to talk instead of type, in any app instead of only Docs, that's dictation, and a dictation app does it better. Wispr Flow runs on Mac, Windows, iPhone, and Android, and its free plan covers 2,000 words a week on desktop. Our Wispr Flow review covers 90 days of daily use, and the AI dictation guide compares the rest.

Method 4. Apple Voice Memos and Notes

On iPhone, the free transcriber is already installed. Voice Memos and Notes both turn speech into text as you record. Apple lists it for iPhone 12 or later in English, Spanish, Portuguese, Italian, French, German, Japanese, Korean, and Chinese, per its Voice Memos guide and Notes guide. The Mac version of Voice Memos shows transcripts too, per Apple's Mac guide.

1. Open Voice Memos and tap the Record button.
2. Record, then tap Done.
3. Tap the recording, tap the More button, and choose Copy Transcript.
4. Paste it into Mail, Notes, or any doc.

Apple's guides describe transcribing what you record in those apps, not files you bring in from elsewhere. So it's the best method when the audio doesn't exist yet. We go deeper on the iPhone version, including copying text out and what to do on older phones, in how to transcribe voice memos on iPhone.

Method 5. Otter's Free Plan

Otter is built for recording meetings live, and its free Basic plan is generous on that front. Otter's pricing page lists 300 transcription minutes a month, a 30-minute cap per conversation, and 5 custom vocabulary terms. File uploads are the tight part. Basic gets 3 audio or video imports per user, for life.

1. Create a free Otter account.
2. Use the import option to upload your audio file.
3. Wait for the transcript, then review the speaker labels.
4. Copy or export the text.

Three imports make Otter fine for a one-off and a poor home for a backlog. It's at its best live. Our Descript vs Otter breakdown covers who should pick which.

Method 6. ElevenLabs Scribe, the Accuracy Pick

When the transcript is what you're delivering, we run it through ElevenLabs Scribe. ElevenLabs bills it as industry-leading transcription accuracy across 99 languages, and it adds speaker labels and word-level timestamps without being asked.

1. Sign up for a free ElevenLabs account.
2. Open the Speech to Text tool.
3. Upload your file, and pick the language or let it detect one.
4. Review the transcript, with speakers already labeled.
5. Export it in the format you need.

Here's the math on free. ElevenLabs' pricing page lists 10,000 credits a month on the free plan, and speech to text costs about 330 credits a minute, so you get roughly 30 minutes a month. The free plan doesn't include a commercial license. That starts with Starter at $6 a month. We map exactly what free gets you in our ElevenLabs free plan breakdown.

Method 7. Descript, When You'll Edit the Audio

If the transcript is a step toward editing a podcast or video, Descript transcribes on import and then lets you edit the media by editing the text. Delete a sentence, and it's cut from the audio. The free plan includes 60 media minutes a month, with 720p exports that carry a watermark. The Hobbyist plan runs $16 a month billed annually with 10 media hours.

1. Create a free Descript account and start a new project.
2. Drag your audio or video file in. Transcription starts on import.
3. Fix any wrong words right in the transcript.
4. Export the transcript, or keep going and edit the media.

If you never touch the audio itself, you're paying for an editor you don't use. In that case, you're better off with one of the first six methods.

How to Get a Cleaner Transcript From Any Tool

The tool matters less than the audio you're feeding it. A few habits fix most errors before they happen.

1. Get the mic close. A phone on the table a foot away beats a laptop mic across the room.
2. Use the original file. Upload the first recording, not a compressed copy pulled from a video site or a chat app.
3. Cut the crosstalk. Two people talking at once is where every model breaks down.
4. Teach it your words. Custom vocabulary, where the tool offers it, saves you from fixing the same name 40 times.
5. Proofread the risky parts. Names, numbers, dollar amounts, and anything you'll quote.

Which One We'd Use

For a pile of long files and zero budget, it's Audacity with Whisper. It's the only option here with no cap on uploaded files. For short jobs where accuracy and speaker labels matter, ElevenLabs. If you're on Microsoft 365 already, try Word before paying for anything.

And if the audio you keep transcribing is your own meetings and calls, stop doing it by hand. A recorder transcribes as it goes. Pocket and Plaud Note both clip onto your phone, and Plaud's free Starter plan covers 300 minutes a month. For YouTube videos specifically, the free transcript panel usually beats all seven, which we cover in how to transcribe a YouTube video.