How to transcribe audio to text, step by step: from MP3 to text with speakers and timestamps
To transcribe audio to text, upload the file to FilmScribe as it is: MP3, WAV, M4A, OGG or OPUS, with no conversion. In minutes you get the text word by word, with speakers separated and timestamps. Review it by listening, fix names and terms, and export TXT, PDF or SRT. The 3-day trial asks for a card.
Step by step
- 01 Upload the audio file In FilmScribe, pick the recording as MP3, WAV, M4A, OGG, OPUS or another accepted format, exactly as it came off the recorder or your phone.
- 02 Set the content type and mark the range Choose the type, like podcast, interview or meeting. If only part matters, mark the range: only that part is processed and counts toward your hours.
- 03 Get the transcript We transcribe word by word, with a timestamp on every word, speakers separated and the language detected automatically.
- 04 Review by listening Click a sentence to hear that passage, and fix names, acronyms and technical terms right in the text.
- 05 Export the right format TXT for documents, PDF for reading, SRT for captions and an MP3 of the passage you marked.
In this guide
- The short version
- How to transcribe audio to text in FilmScribe
- What changes when the file is audio only?
- Which audio formats can you transcribe? MP3, WAV, M4A, OGG and OPUS
- How do you tell who spoke, and when?
- Which format should you export: TXT, PDF or SRT?
- How much does audio quality affect the transcript?
- How do you transcribe a long audio file, over an hour?
- How do you transcribe audio on your phone?
- Audio in another language: transcribe or translate?
- Is transcribing audio the same as speech to text?
- Can you transcribe audio to text for free? The options and their limits
- Common mistakes when transcribing audio
- Frequently asked questions
- Sources
Transcribing audio to text means getting everything said in a recording onto the page: the interview sitting on your recorder, the podcast episode, the meeting, the lecture, the voice note. The file might be an MP3, a WAV, an M4A, an OGG or an OPUS. The job is the same every time: hear each word, know who said it and note when.
In FilmScribe, you upload the file as it is and we return the transcript word by word, with a timestamp on every word and speakers separated. This guide walks through the steps, what changes when the file is audio only, what to know about each format, how to export and what recording quality does to the text. If you just want to upload a file now, start at the free audio to text page: it takes you to the 3-day trial, which asks for a card.
The short version
- Upload the audio as it is: MP3, WAV, M4A, AAC, OGG, OPUS or WEBM, up to 10 GB, with no conversion first.
- We transcribe word by word, with timestamps, speakers separated and the language detected automatically, among 90+ languages.
- An audio file gets the transcript, chapters and emotions on every plan, plus hooks and scripts on Studio and Agency. It doesn’t get the Clips page or burned-in captions, which need video.
- Recording quality matters more than file format. Review by listening before you publish.
- Export for the destination: TXT for documents, PDF for reading, SRT for captions.
How to transcribe audio to text in FilmScribe
1. Upload the audio file
Pick the recording in the upload area. MP3, WAV, M4A, AAC, OGG, OPUS and WEBM all go in exactly as they came off the recorder, the phone or the messaging app. Each file can be up to 10 GB and there’s no length limit; what counts is your plan’s hours. See the supported formats.
2. Set the content type and mark the range
Choose the content type, such as podcast, interview, meeting, class or narration. It guides what we look for in the recording. If only part matters, mark the start and end before processing: only that range is transcribed and only it counts toward your hours.
3. Get a word-by-word transcript
We transcribe with a timestamp on every word and separate the lines by person, on every plan. The language is detected on its own. Once the upload finishes you can close the page: we email you when the file is ready. See transcription and speakers.
4. Review by listening
As the audio plays, the word being spoken is highlighted. Click any sentence and the audio jumps to it. A name, an acronym, a technical term: if something came out different from what was said, fix it in place. The correction applies to the text on screen and to every export, and you can always go back to the original. Search any word in the file, filter by person and star the lines you want to keep.
5. Export
TXT and PDF to read, SRT for captions with TV, Netflix and YouTube presets, and an MP3 of the whole file or just the passage you marked (download snippets). The export lets you choose whether speaker names and times go into the text.
What changes when the file is audio only?
The transcript is the same one a video gets. What changes is what comes with it:
| What you’re after | Audio file | Video file |
|---|---|---|
| Transcript with timestamps and speakers | Yes, on every plan | Yes, on every plan |
| Chapters and emotions | Yes, on every plan | Yes, on every plan |
| Hooks tab | Yes, on Studio and Agency | No: the automatic clips play that role |
| Scripts | Yes, on Studio and Agency | Yes, on Studio and Agency |
| Clips page, with vertical clips | No | Yes |
| Burned-in captions | No, there’s no picture to burn on | Yes, on clips |
| SRT caption file | Yes | Yes |
| Download of one passage | MP3 | MP3 or MP4 |
| Cut timeline for your editor | Yes, on Studio and Agency | Yes, on Studio and Agency |
Three things worth knowing:
- The Hooks tab only exists for audio. On Studio and Agency, it lists the lines that grab attention, each with a 0 to 100 score, a type, a reason and timestamps. It’s how you find the strong moment in a podcast without listening to the whole thing again. See narrative hooks.
- Emotions come from what was said. We read the transcribed text, not the tone of voice. A stretch with no speech gets no emotion.
- A vertical clip needs a picture. If the recording exists on video, upload the video: the transcript is the same and you get the Clips page too. Our guide on how to transcribe video to text covers that workflow.
Which audio formats can you transcribe? MP3, WAV, M4A, OGG and OPUS
Every common format goes in without conversion. What differs from one to the next is where it comes from and how big the file is:
| Format | Where it usually comes from | What to know |
|---|---|---|
| MP3 | Podcasts, handheld recorders, downloads | Compressed and light: an hour at 128 kbps is about 58 MB |
| WAV | Pro recorders, mixing boards, studios | Uncompressed: an hour of CD-quality stereo runs past 600 MB |
| M4A | Phone recorder apps, meeting recordings | Compressed, good quality in little space |
| OGG | Voice notes, some recorder apps | Compressed; not every player opens it, but we do |
| OPUS | Voice notes from messaging apps | Built for speech and music over the internet, from 6 kbit/s up (RFC 6716) |
| AAC, WEBM | Online recorders, some apps | Accepted as they are |
Each format has its own upload page if that’s what you searched for: MP3 to text, WAV to text, M4A to text and OGG to text. For voice notes, our guide on how to transcribe WhatsApp audio shows how to get the file off your phone.
Three practical rules about format:
- Don’t convert before uploading. Turning an MP3 into a WAV only makes the file bigger. What compression removed doesn’t come back.
- FLAC isn’t on the list. Run it through the free FLAC to MP3 converter and upload the MP3.
- If the sound lives inside a video, upload the video. We transcribe its audio the same way. It’s only worth using extract audio from video first when the video is too heavy for your connection, knowing that an audio file has no clips.
How do you tell who spoke, and when?
They’re two separate labels, and both come built in.
Speakers. Every line arrives labeled with who said it: Person 1, Person 2 and so on, on every plan. You can filter the transcript by person to read only the guest’s answers, for example. On Studio and Agency you can give each person a name and a color, or ask us to work out who is who: we find the moments where people introduce themselves or get called by name and suggest a name for each, with the passage that proves it. The name you pick shows up in the transcript, the filters and the exports. One honest limit: you can’t split one speaker into two.
Timestamps. Every word carries its own timestamp, not just every sentence. Click the time on any line and the player goes there. That’s what lets you quote precisely (“at 14 minutes 32 seconds”) and download only the answer you need as an MP3.
An audio transcript with speakers and timestamps looks like this:
[00:14:32] Host: What was the most expensive mistake of your first year?
[00:14:36] Ray: Hiring before I had a process. I brought on three people
and none of them knew what to do on Monday morning.
[00:14:45] Host: And how did you fix it?
For research interviews and their citation rules, see how to transcribe an interview.
Which format should you export: TXT, PDF or SRT?
Pick by where the text is going:
| Format | What it’s for | Keeps timing? |
|---|---|---|
| TXT | Pasting into a document, opening in Word, editing | Can include timestamps in text |
| Sending to a reader, printing, archiving | Can include timestamps in text | |
| SRT | Captions for a video made from this audio | Yes, in and out time per line |
| MP3 | The whole audio or just the marked passage | Not applicable |
When you export, you decide whether speaker names and times are included, the caption line width (TV, Netflix, YouTube or custom) and the frame rate.
What doesn’t exist, so you don’t go looking: DOCX, VTT, JSON or CSV export, and WAV audio. Audio always comes out as MP3. The workarounds:
- Need a .docx? Open the TXT in Word and save it as a document. Our guide on how to transcribe audio in Word explains Word’s own Transcribe feature and its limits.
- Need VTT? The free subtitle converter turns the SRT into VTT without reprocessing.
- Only want captions from the audio? The MP3 to SRT page leads to exactly that, and our guide on how to create an SRT file explains what’s inside the file.
The same passage from the example above, as SRT:
41
00:14:36,100 --> 00:14:39,400
Hiring before I had a process.
42
00:14:39,550 --> 00:14:44,800
I brought on three people and none of them
knew what to do on Monday morning.
How much does audio quality affect the transcript?
A lot. The accuracy of automatic transcription depends on the recording far more than on the file format:
| Recording problem | What happens in the text | How to fix it |
|---|---|---|
| Phone in the middle of the table | Swapped words, quiet lines dropped | Move the recorder closer or use a lavalier mic |
| Echoey room | Syllables drop or repeat | Record in a smaller room with curtains or a rug |
| Music, TV or street noise behind | Whole passages skipped | Turn the noise off or move somewhere else |
| Two people talking at once | Mixed lines, wrong speaker | Review those passages by listening again |
| Phone call or speakerphone | Muffled voice, more wrong words | Record on the speaker’s own device when you can |
| Clipped, too-loud signal | Distorted words | Back the mic off or lower the gain |
| Names, acronyms and technical terms | Wrong spelling or a sound-alike word | Fix in the text; the fix carries over |
A quick test: listen to one minute of the recording on headphones. If you catch everything without rewinding, the transcript will need few fixes. If you have to replay a line to get it, that’s where your review time will go.
Before a recording that matters, record ten seconds and listen back. The free mic test shows your level, peak and noise floor. For the whole process, from recording to text, see how to record and transcribe audio.
Even with clean audio, review it. The W3C’s guidance on transcripts says to plan time for correcting automatically generated text, because a single missing word, such as a “not”, can flip the meaning of a sentence.
Verbatim or clean?
A verbatim transcript records everything: “um”, “you know”, repetitions, laughter. According to the W3C, that’s what legal depositions call for. A clean transcript drops the filler and keeps the meaning, and it’s the usual choice for reading, quoting and publishing. Decide before you review. If the destination is running text, the free transcript cleaner strips timestamps and speaker labels and joins the lines into paragraphs.
How do you transcribe a long audio file, over an hour?
Upload it as it is. There’s no length limit: the limit is size, 10 GB per file, and what counts is your plan’s hours, from 10 to 35 a month.
- Only part of it matters? Mark the range before processing. Only that range is transcribed and only it counts toward your hours.
- On the trial? The trial includes 1 hour of material. A longer file is transcribed up to the hour, and the rest is kept to be transcribed after you subscribe, with no second upload.
- The recording came as several files? On Agency you upload up to 20 audio files at once and choose between one transcript each or everything joined into one. On the other plans, upload them one at a time and keep them in a project, or merge the parts first with the free audio joiner.
- No need to wait around. Once the upload finishes, close the page. We’ll email you when the file is ready.
Nobody reads a two-hour transcript top to bottom. Use search to jump to a topic, chapters to see the recording split by subject and the person filter to read one side of the conversation.
How do you transcribe audio on your phone?
Open FilmScribe in your phone’s browser and pick the file saved on the device. The process is the same as on a computer.
Two things to watch when uploading from a phone. Keep the page open until the upload finishes: switching apps or locking the screen pauses it, and it picks up where it stopped when you come back. And use Wi-Fi for big files, like an hour-long WAV.
Reviewing a long transcript is easier on a bigger screen. A routine that works: upload from your phone right after recording, then review at your desk.
Audio in another language: transcribe or translate?
Transcribe first, in the language that was spoken. The language is detected automatically, among 90+ languages, and the transcript comes out in that language. With the original reviewed, the translation is more faithful: translating straight from sound stacks a hearing error on top of a translation error.
On Studio and Agency, we translate the transcript into 9 languages, with the same timing as the original. Recordings that switch languages need extra care where they switch. Our guide on how to translate audio covers the whole process.
Is transcribing audio the same as speech to text?
People use “speech to text” for two different jobs, and the difference decides which one you need.
Transcribing starts from a recording that already exists. You upload the file and get the full text, with who spoke and when. It fits interviews, meetings, classes, podcasts and any conversation with more than one person.
Voice typing is dictation: you talk into the mic and the text appears as you speak, with no file, no speakers and no timestamps. It fits writing a message, a note or a first draft. Our free speech to text tool does that, and the guide on how to use voice typing covers it step by step.
What doesn’t work well is mixing the two: playing a recording out loud next to the mic so dictation can “hear” it. The room gets into the text, voices blend and punctuation disappears.
Can you transcribe audio to text for free? The options and their limits
FilmScribe has no free plan. What it has is a trial, and other tools handle the simple cases. Each comes with its own limit:
| Option | What you get | Limit |
|---|---|---|
| FilmScribe trial | Your file transcribed, with speakers, timestamps and exports | 3 days or 1 hour of material; asks for a card |
| Voice Memos on iPhone | A transcript of recordings made in the app, to view and copy | iPhone 12 or later, in some languages (English among them), and not in every country or region |
| Voice typing in Google Docs | What you say into the mic, typed into the document | Uses your computer’s microphone, not a file; current Chrome, Edge and Safari |
| Transcribe in Word | An uploaded file, with speakers separated | Microsoft 365 subscription; 300 minutes of uploaded audio a month; .wav, .mp4, .m4a, .mp3 only |
| Typing it by hand | Any recording, as carefully as you like | Takes several times the length of the audio |
On the FilmScribe trial, if you cancel in the app before the 3 days are up, nothing is charged. If you don’t cancel, the subscription to the plan you picked begins. The free audio to text page shows what the trial delivers and takes you to the upload.
For a two-minute memo, your phone’s recorder is enough. For an hour-long interview, a meeting with four voices or a voice note in OPUS, a format Word doesn’t currently take, what matters is review time: text with speakers separated and a time on every word reviews far faster than one long block.
Common mistakes when transcribing audio
- Skipping review. No automatic transcript should go straight to publication. One read while listening catches most errors.
- Converting the file before uploading. Changing the format doesn’t improve the recording and costs you time. Upload the original.
- Uploading the audio when the video exists. The text is the same, but only video gets vertical clips and burned-in captions.
- Recording with the phone far away. Three feet less between mouth and mic beats any fix you make afterward.
- Exporting without timestamps or speakers. The text seems enough until you need to find a line in the recording or know who said it.
- Playing the recording for dictation to hear. Dictation is for your own voice, live. A finished recording gets transcribed from the file.
Frequently asked questions
Can I transcribe audio to text for free?
FilmScribe has no free plan. The 3-day trial, with 1 hour of material and a card on file, transcribes your own audio with speakers, timestamps and TXT, PDF and SRT export; cancel before it ends and you aren't charged. Without paying anything, the recorder app on some phones transcribes its own recordings as plain text.
How do I transcribe a long audio file, over an hour?
Upload the whole file. There's no length limit, only a size limit (10 GB per file), and what counts is your plan's hours. If only part matters, mark the range before processing. On the trial, a file longer than 1 hour is transcribed up to the hour, and the rest is kept for after you subscribe.
How do I convert an MP3 or another audio file to text?
Upload it as it is. MP3, WAV, M4A, AAC, OGG, OPUS and WEBM go in without conversion, which covers phone recorders, field recorders, mixing boards and voice notes. FLAC isn't on the list, so convert it to MP3 first. Turning an MP3 into a WAV won't improve the text: what compression removed doesn't come back.
Can I transcribe audio with timestamps and speaker labels?
Yes. We timestamp every word and separate the lines by person on every plan, as Person 1, Person 2 and so on. Click a sentence and the player jumps to that exact point in the recording. Naming each speaker and asking us to work out who is who are on Studio and Agency.
Is transcribing audio the same as speech to text?
Not always, because speech to text covers two jobs. Transcribing starts from a finished recording: you upload the file and get the full text with speakers and timestamps. Voice typing is dictation: you talk into the mic and the words appear live, with no file, speakers or timing. Transcribe recorded interviews, meetings and podcasts; dictate to write a note.
Which format should I export the audio transcript in?
TXT to paste into a document or open in Word, PDF for someone who only needs to read it, and SRT when the text becomes captions on a video made from that audio. The SRT has TV, Netflix and YouTube presets. We don't export DOCX or VTT: open the TXT in Word, or convert the SRT.
Can I make clips or captions from an audio file?
Vertical clips and burned-in captions need a picture, so they don't exist for audio files. What audio gets is the transcript, chapters, emotions and MP3 downloads of any passage; on Studio and Agency, also the Hooks tab and scripts. The SRT caption file exports as usual. If the recording exists on video, upload the video.
Sources
The official documentation behind the numbers and rules in this guide.
- 01Transcribing Audio to Text(opens in a new tab) · W3C WAI
- 02Transcribe your recordings(opens in a new tab) · Microsoft
- 03Type & edit with your voice(opens in a new tab) · Google Docs Help
- 04View a Voice Memos transcription on iPhone(opens in a new tab) · Apple
- 05RFC 6716: Definition of the Opus Audio Codec(opens in a new tab) · IETF