4.4
2,100+ reviews on G2
Transcribing audio means turning the speech in a recording into written text, either by typing it as you listen or with speech-to-text software that you then check. A usable transcript has four things:

What was said, as said, not a summary.
A speaker label on each turn when more than one person talks.
So you can find any line in the recording again.
Names, numbers and unclear words verified against the audio.
Transcribe it in Evernote. Drop in a recording and get a transcript with speakers, then ask for any quote.
Use the original recording, not a copy of a copy. If you are still recording, put the microphone close to the speaker and away from noise.
Type it yourself for short, sensitive or very noisy audio. For anything longer, let speech-to-text software do a first draft, then check it.
In Evernote, add the file to a note and choose to transcribe it. Evernote accepts common audio files such as MP3, M4A and WAV.
For interviews and meetings, ask the tool to identify who said what. Rename the speakers once, and the names carry through the transcript.
A timestamp at the start of each turn lets you jump back to check a line or pull a quote.
Listen again wherever the text looks wrong. Mark what you can't make out, for example [inaudible 12:05].
A basic transcript includes non-speech audio needed to understand the content, such as [laughter] or [phone rings].
| Time | Speaker | What was said |
|---|---|---|
| 402:10 | You | How did the funding cut change the library? |
| 514:52 | Ana Rivera | We lost a third of our budget in one vote, and the first thing to go was Sunday hours. |
| 617:40 | Ana Rivera | The council voted 5 to 2. [inaudible 17:48] was the only one who asked about the kids. |
| 721:30 | Ana Rivera | Kids were doing homework in the parking lot, just to use our Wi-Fi. [laughs] |
Speaker names, timestamps at the start of each turn, unclear words marked with the time.
Each option trades time against accuracy and cost. Checks are on each maker's own help pages, November 2026.
| Method | What you do | Good to know |
|---|---|---|
| By hand | Listen, pause and type | Free and private, but slow |
| iPhone Voice Memos | Record, then view or copy the transcript | iPhone 12 or later, 10 languages |
| Word for Microsoft 365 | Upload a WAV, MP4, M4A or MP3 file to Transcribe | Windows, commercial; 300 min a month |
| Evernote | Add the file to a note and transcribe it | Speaker labels, timestamps, language detected or chosen |
| ChatGPT | Attach an audio file and ask for a transcript | Audio uploads need a paid ChatGPT plan; up to 6 minutes per pass |
Sources: Apple Support, Microsoft Support, OpenAI Help Center.
Copy it, or save it to your Evernote account and use it for every transcript you clean up.
Recording: [file name], [date], [length]
Speakers: [name] (host), [name] (guest)
[00:00] [Speaker]: [what was said, word for word]
Unclear: [inaudible mm:ss] or [crosstalk mm:ss]
Sounds that matter: [laughter], [music], [phone rings]
Checked by [name] on [date]: names, numbers, quotes
4.4
2,100+ reviews on G2
4.4
8,200+ reviews on Capterra
4.4
73,000+ reviews on App Store
248M
Registered users
Type it yourself, or use a tool already on your device. iPhone Voice Memos shows a transcript of recordings on iPhone 12 or later, and Google Docs voice typing turns live speech into text in Chrome, Edge or Safari.
Longer than the recording itself, because you pause and rewind. You can let software make a first draft and spend your time checking it.
Yes. Word for Microsoft 365 on Windows in commercial tenants can transcribe uploaded WAV, MP4, M4A and MP3 files, up to 300 minutes a month with a Microsoft 365 subscription.
Yes. You can attach an audio file and ask for a transcript; audio uploads are available on paid ChatGPT plans, up to 6 minutes per pass. OpenAI notes that transcription can contain mistakes, so check important parts.
Use a tool that labels speakers, such as Evernote's "Identify who said what", then rename the speakers. Keep a timestamp at the start of each turn so you can check quotes.
Captions are timed text that shows on the video as people speak. A transcript is the full text, read separately, and can include speaker names and sounds.
Rules on recording other people vary by place. Ask for consent before you record, and check the law where you and they are.
It depends on the audio. Check names, numbers and anything you will quote against the recording.
Put the recording in a note, and Evernote writes the transcript under it, with who said what.
Add an audio or video file to a note and choose to transcribe it. Turn on 'Identify who said what' to get speaker labels and timestamps.
Ask the AI Assistant where a topic comes up and for the exact words. It reads the transcript and answers with the timestamp.
Ask the Assistant to summarize the recording, for example as key points, and the summary appears above the transcript.
Drop in the file, get a transcript with who said what, then ask the AI Assistant for the quote or the key points.
Start for free