QuillHubQuillHub
RU

MP3 to text converter: free transcription online

Upload an MP3 and QuillHub turns the recording into text: punctuation, paragraphs, speaker labels, and timestamps. Files up to 10 hours, no converting or installing.

20,000+hours transcribed

60 free minutes when you sign up · No card required

  • Google Drive
  • Direct file link
  • Files MP3, WAV, M4A, OGG, AAC, WMA

Sample transcript

What an MP3 looks like as text

Real QuillHub output with no manual edits: 46 minutes of recording, 5 voices separated automatically, every line with its own timestamp. Above the text: chapters and a summary. Switch the tabs to see each layer.

Telling Time on Other Worlds — NASA JSC Podcast Ep.41900:00 / 45:37 · Kevin Coggins, Dane Turner, Podcast promo, Mission audio, Archival audioReady · 4 documents
  • Dane Turner0:36

    Houston, we have a podcast. Welcome to the official podcast of the NASA Johnson Space Center, episode 419, Telling Time on Other Worlds. I'm Dane Turner, and I'll be your host today.

  • Dane Turner8:57

    And so that's what becomes 1 second. And then we extrapolate all of the other times from that second?

  • Kevin Coggins13:23

    Well, it could have been a watch on someone's wrist, right? It could be a clock in one of the— in the capsule, or, you know, a clock in some other part of the system.

  • Dane Turner19:46

    So are gravity and rotation the only things that we really need to keep into consideration when we're building a time for another planet?

  • Dane Turner30:29

    So one of the things that you said just a minute ago was that these clocks will have to endure temperature. Is temperature something that affects time?

  • Dane Turner34:20

    So in 2024, SCaN was tasked by the White House with developing this coordinated lunar time. Uh, how is that project coming along?

  • Dane Turner41:21

    Right. Okay. So the people working it were on the day-night cycle and not living on the Martian clock. Okay. So is, is Mars our next focus after we get the Moon infrastructure up?

89 more turns, each with a speaker label and a timestamp; click a timestamp to jump the audio there

Speech split by person

Free speech-to-text hands back one paragraph for the whole recording. Here the voices are separated and every turn carries its own timestamp, so the text is readable straight away.

  • 5 voices separated automatically, no setup
  • Kevin Coggins — 71%, Dane Turner — 27%, Podcast promo — 2% of the time
  • Export to TXT, DOCX, SRT, and VTT
  • 0:00Introduction and Episode Overview2 min
  • 2:06Guest Background and SCaN (Space Communications & Navigation) Overview1 min
  • 3:35Lunar Navigation and the Challenge of Timekeeping2 min
  • 5:35What Is Time? Atomic Clocks, Gravity, and Reference Frames3 min
  • 8:37Atomic Clocks and Precise Timekeeping3 min
  • 11:19Moon Reference Frames and the Need for Lunar Time4 min
  • 15:02Mission Elapsed Time, UTC, and Transition to Lunar Time3 min
  • 17:45Gravity, Rotation, and Human Factors in Planetary Timekeeping3 min
  • 20:34Lunar Timekeeping Challenges2 min
  • 22:35Defining Lunar Coordinate Systems3 min
  • 25:06Moon vs Earth: Physical Differences1 min
  • 25:46Clock Placement and Synchronization2 min
  • 27:58Operational Errors and Safety Use Cases2 min
  • 30:27Clock Stability and Temperature Effects2 min

9 more chapters, each with its own timestamp and length

The whole 46 minutes at a glance

Chapters follow the meaning, not the pauses, so you can see what each part of the recording covers and jump to the one you need.

  • 23 chapters across 46 minutes
  • Each chapter links to its place in the recording

Interview with Kevin Coggins (Deputy Associate Administrator, SCaN) about challenges and plans for precise timekeeping off Earth.

  • Gravity and orbital dynamics change atomic clock rates, so "time" differs between Earth, Moon, and Mars.
  • Precise timekeeping (nanosecond accuracy) is essential for accurate positioning and safe operations on other bodies.
  • A coordinated lunar time reference frame will be required for multi-actor activities on the Moon; mission elapsed time is insufficient for a sustained presence.

Decisions the AI pulled out

  • White House tasked SCaN (in 2024) to develop a coordinated lunar time.01
  • Artemis II will use Mission Elapsed Time as its primary operational timeline reference starting at launch.02

No need to replay 46 minutes

The summary is built from the broadcast itself: key ideas, numbers and the decisions that were actually announced. It can go into a team chat as it is.

  • 2 decisions pulled from the conversation
  • 6 key points in the full summary
One-page digest

SCaN is developing a lunar communications and navigation relay system (LCRNS/LCORNS) to provide GPS-like services at the Moon.

from the recording
Article draft

“Telling Time on Other Worlds — NASA JSC Podcast Ep.419”: lunar time, atomic clocks, space PNT

from the recording
Follow-up email

What was discussed, what was decided, what happens next

template
Task list

Who does what next — the tasks from the recording as one list

template
Ask the recording

“Create a checklist for deploying lunar PNT infrastructure (clocks, relays, surface nodes)”: the chat answers from the text and gives the timestamp

chat
Your own prompt

A CRM note, a post, talking points — or any format you need

anything
  • SCaN
  • PNT
  • LCRNS / LCORNS
  • Deep Space Network
  • UTC
  • MET
  • Sol

Every AI action is included in the Unlimited plan — no credits per click.

Finished documents on top of the text

The transcript is only the start. From one recording you can build a digest, an article draft or an email — or describe a format of your own.

  • Ready-made templates and your own prompts
  • Chat over the recording: ask anything about the text
  • 7 terms with definitions
Upload your recordingSee the full example(opens in a new tab)60 free minutes when you sign up · No card required

Example: “Telling Time on Other Worlds,” Houston We Have a Podcast, episode 419, NASA, license public domain (U.S. government work). Transcript, chapters, summary and key points by QuillHub.

How it works

How to convert MP3 to text

Three steps in your browser. Nothing to install.

  1. 1

    Upload your MP3

    Drag the file into the form or pick it from your computer. You can add several files at once, each up to 10 hours. If the recording is in Google Drive, paste the link instead. Then press Start transcription.

  2. 2

    Wait for the transcript

    QuillHub recognizes the speech, adds punctuation, and splits the text into paragraphs and speakers. The language is detected automatically, and you can set it under Speaker Recognition & More Settings.

  3. 3

    Download the text

    The finished transcript opens in your browser. Export it as DOCX for Word, TXT, or Markdown, or as SRT and VTT subtitles.

What you get

MP3 to transcript you can send right away

Raw speech recognition gives you one long stream of words. Here the recording becomes a document you can read, search, and quote.

Punctuation and paragraphs

Periods, commas, and question marks are added automatically, and a long recording is broken into paragraphs.

Speaker labels

In an interview, call, or podcast, voices are separated automatically. Rename Speaker A and Speaker B to real names once, and the names carry into the export.

Timestamps

In recordings longer than three minutes, every line has a time mark. Click it and the audio jumps to that moment, so you can recheck a quote or a number.

Word, TXT, and Markdown

Download the transcript as a DOCX file for editing in Word, as plain TXT, or as Markdown for notes and docs.

SRT and VTT subtitles

If the MP3 is the soundtrack of a video, export the transcript as timed subtitles.

Summary and chapters

For recordings longer than three minutes, QuillHub writes a summary, key points, and chapters with timestamps.

AI on top of the transcript

MP3 to text, then answers from the recording

Once the MP3 is text, ask the recording a question and get the answer with the moment it was said.

User interview · 2 speakers38:12
What made her stop using the old app?

Exports broke every time a recording ran past an hour, and she had to split files by hand. She said she'd switch for that alone.

Interview · 21:47 — the fragment is lit on the track
Did other interviewees mention long files?

Yes, two of the five. One records two-hour lectures, the other exports weekly team calls.

5 interviews in the archive
interview 2 · lectures
interview 4 · team calls
01

Ask across all recordings

Ask the chat about every interview or call at once: what people repeated, what they asked for.

02

Follow-up, tasks, blog post

One click for a summary, a follow-up email, a task list, or a blog post draft. Or write your own prompt.

03

Access for your AI agents

The MCP server lets Claude and other AI agents work with your recordings directly.

Every AI action is included in the Unlimited plan — no credits per click.

Ways to do it

Four ways to turn an MP3 into text

Google and Microsoft both have speech-to-text built in, but only one of them takes a file. Here is how the options compare for a recorded MP3.

OptionHow it handles an MP3Speakers and timestampsLimitsGood for
QuillHubUpload the file or paste a Google Drive linkSpeakers on by default, a timestamp on every line60 free minutes, files up to 10 hours and 5 GBLong recordings where you also want a summary and chapters
Microsoft Word TranscribeUpload an .mp3, .wav, .m4a, or .mp4 in Word for Microsoft 365Speakers separated300 minutes of uploaded audio a month with a Microsoft 365 subscriptionYou already pay for Microsoft 365 and want the text in Word
Google Docs voice typingListens to your microphone; there is no file upload, so the MP3 has to be played out loudNot described in Google's helpChrome, Edge, or Safari; the MP3 plays in real timeDictating short notes in your own voice
Typing it yourselfPlay, pause, type, repeatWhatever you add by handYour timeA clip of a minute or two, or a final check of a quote

Which MP3 files people convert to text

MP3 is the format most recorders, phone systems, and podcast hosts save to, so the jobs vary a lot.

Podcast episodes

An episode MP3 becomes a full transcript for the show page, plus chapters with timestamps for the episode description.

Interviews

A recorded interview becomes text split by speaker. You edit and quote instead of transcribing by hand.

Lectures and webinars

A lecture recording becomes text with chapters, so you can find a topic by the outline instead of scrubbing through audio.

Sales and support calls

Phone systems and CRMs often save calls as MP3. Speaker labels show what the rep promised and what the customer said.

Meetings

An MP3 of a team call gives you the text, a summary, and the decisions, with no one taking minutes.

Voice notes

Ideas recorded on the go turn into text you can edit, paste into a doc, or send.

Who it's for

Who saves the most time with MP3 transcription

The more recordings pile up each week, the bigger the difference.

Podcasters

A transcript and show notes for every episode

Upload the episode MP3 and get the full text, chapters with timestamps for the description, and a first draft of a blog post.

  • Chapters with timestamps for show notes
  • Blog post draft from the episode in the AI chat
  • SRT subtitles for the video version
QuillHub for media

Journalists and researchers

Interviews ready for editing

An hour-long interview MP3 becomes text split by speaker in a few minutes. Every quote comes with a timestamp to check it against the audio.

  • Lines split by voice
  • A timestamp on every quote
  • DOCX export for your editor
QuillHub for journalists

Sales teams

Call recordings you can read

Export call MP3s from your phone system and get the text of each conversation. Commitments and objections are visible without listening back.

  • Several call recordings in one upload
  • Rep and customer lines kept apart
  • Follow-up email and task list from the AI chat
QuillHub for sales teams

Product and UX teams

Customer interviews you can search

Upload interview recordings and ask the chat what came up across all of them, with the moment each answer was said.

  • Questions across every interview in the archive
  • Answers with timestamps
  • Summary and key points per interview
QuillHub for product teams

Tips

How to get an accurate transcript from an MP3

Accuracy depends on how the audio was recorded, not on the file format. Here is what to check.

  • Don't convert the file first

    MP3, WAV, M4A, OGG, AAC, and WMA upload as they are. An extra conversion only loses audio quality.

  • Clean audio beats a high bitrate

    Background noise, echo, and people talking over each other hurt accuracy the most. A regular MP3 from a recorder or a phone system is fine for speech recognition.

  • Set the language for heavy accents

    The language is detected automatically, one per recording. If the speech has a strong accent, pick the language under Audio Language before you start.

  • Split files with two languages

    A recording is transcribed in its main language. If two languages take turns half and half, cut the MP3 into parts and upload each with its own language.

  • Keep long recordings whole

    A file up to 10 hours is processed in one go. Cutting it up breaks the timestamps and the chapters.

  • Fix a misheard name once

    If a name or term comes out wrong, open Edit text and use Find and replace. The fix applies to the text, the chapters, and the summary.

Without the website

MP3 to text in Telegram

If you use Telegram, you can also send an MP3 to @QuillAI_Bot. It is the same service, and once you link Telegram to your account, the transcripts from the bot show up in the web app too.

Open @QuillAI_Bot
  • Send the file from any chat or from your phone
  • Minutes and plan are shared with the website after you link Telegram
  • After linking, transcripts from the bot land in your account

Pricing

The more you record, the cheaper the hour

Start free, pay only when you need more.

Our lowest price
Unlimited

from$6.66per month

Six months of unlimited — $39.99 up front. Minutes are not counted at all.

Get Unlimited

Six months — $39.99, billed once.

What $6.66 a month works out to

10 hours$0.67/hr
20 hours$0.33/hr
50 hours$0.13/hr

The same $6.66 in one-off packs buys about 500 minutes — roughly eight hours. Then they run out; Unlimited doesn’t.

Month by month the same half-year would cost $59.94 — a $19.95 difference.

One-off minute packs
from $2.99
Not a subscription. Minutes never expire and stack with your balance.
150 minutes — $2.99350 minutes — $4.99750 minutes — $9.992,950 minutes — $29.99

The minute gets cheaper with volume: from $0.020 in the smallest pack to $0.010 in the largest.

Start
$0
60 minutes when you sign up
  • 60 minutes to try, they don't expire
  • All languages and speaker recognition
  • Export to a document
  • No card required
Get started
Team

A shared workspace, projects and roles. Invoice billing and paperwork for companies.

$14.99per seat, per month
Start a team

FAQ

MP3 to text: frequently asked questions

If your question isn't here, contact support.

MP3 to text is converting the speech in an MP3 audio file into written text. A good converter also adds punctuation, splits the text by speaker, and marks timestamps, so the result reads like a document rather than a stream of words.

First a speech recognition model turns the audio into words and tells the voices apart. Then QuillHub adds punctuation, paragraphs, speaker labels, and timestamps. For recordings longer than three minutes, a language model writes a summary, key points, and chapters on top of the transcript.

Yes. You get 60 free minutes when you sign up, and no card is required. That covers an hour-long recording or several short ones. After that, you can buy a minute pack from $2.99 or switch to Unlimited for $9.99 a month.

Not from a file. Google Docs voice typing listens to your microphone, and Google's help doesn't describe a way to upload audio, so you would have to play the MP3 out loud. Microsoft Word for Microsoft 365 can transcribe an uploaded MP3, up to 300 minutes a month. QuillHub takes the file directly and gives you 60 free minutes to start.

Up to 10 hours and 5 GB per file. An hour of MP3 at a typical 128 kbps is about 58 MB, so even a 10-hour recording fits with room to spare.

Usually a few minutes. In our test, a 46-minute podcast MP3 was ready with chapters and a summary in under 5 minutes. Once you press Start transcription, you can close the page; the transcript will wait in your account.

Yes. Speaker recognition is on by default on every plan. You can rename speakers, and the names appear in the transcript and in the export. Very similar voices can end up under one label.

Yes. Export the transcript as DOCX for Word, or as TXT or Markdown. For video and editing, there are SRT and VTT subtitles. Before downloading, you choose what goes into the file: the summary, the text by chapters with speaker names, or the text with timestamps.

Accuracy depends on the audio more than on the format. Noise, echo, crosstalk, and rare names lower it. Names and numbers are worth checking against the audio, which is one click away on any timestamp.

QuillHub recognizes speech in 98+ languages, one language per recording. The transcript is in the original language; the Translate button in the recording window translates it into one of 11 languages. Translation doesn't use up minutes, but it needs minutes on your balance or an active plan.

No. MP3 to text works in your browser on Mac and Windows. If you also want to record calls on your computer, QuillHub Desktop for macOS and Windows records them and can import audio files too.

Yes. The QuillHub REST API takes a direct file URL or an uploaded file and returns the text, speaker-labeled segments, a summary, and SRT and VTT subtitles. There is also an MCP server for Claude, Cursor, and other AI assistants.

Only you, and anyone you send a share link to. Share links are off until you turn them on. Files and transcripts are encrypted in transit and at rest.

Your data

What happens to your recordings

Doctor's appointments, legal consultations and therapy sessions go through QuillHub. That is why we say this plainly instead of burying it in the terms.

Who processes your data

Encryption — files and transcripts are encrypted in transit and at rest, and only you and the people you share a link with have access.

Never used for training — recordings are not used to train models and are not passed to third parties.

Delete anytime — a recording can be deleted at any moment, together with its transcript and everything derived from it.

Transparent subprocessors — the list of services involved in processing is published on its own page.

Convert your MP3 to text for free

60 minutes when you sign up, no card required. Upload a file, and the text is ready in a few minutes.