QuillHubQuillHub
RU

Interview transcription software for research

Upload an interview recorded on a phone, a voice recorder, or Zoom. QuillHub splits the questions and answers by speaker, adds timestamps, and writes a summary. Download the transcript for Word, NVivo, or your notes.

20,000+hours transcribed

60 free minutes when you sign up · No card required

  • Google Drive
  • YouTube
  • Direct file link
  • Files MP3, M4A, WAV, OGG, MP4, MOV, WEBM

Sample transcript

What a transcript of a 30-minute oral history interview looks like

Real QuillHub output with no manual edits: 30 minutes of recording, 2 voices separated automatically, every line with its own timestamp. Above the text: chapters and a summary. Switch the tabs to see each layer.

Oral History: Richard Gibbs on Early Human Genome Sequencing00:00 / 29:47 · Richard Gibbs, InterviewerReady · 4 documents
  • Richard Gibbs0:13

    I'm Richard Gibbs, and I'm the director of the Baylor College of Medicine Human Genome Sequencing Center at the Baylor College of Medicine in the Texas Medical Center in Houston, Texas.

  • Interviewer0:26

    When and where were you born?

  • Interviewer0:58

    What brought you to the United States?

  • Richard Gibbs1:00

    I did come to Houston and that's where I've been and done my bulk of my professional career. I came as a postdoctoral fellow to pursue general opportunity in science.

  • Interviewer1:46

    When did you first become aware of the Human Genome Project?

  • Richard Gibbs1:50

    Well, I arrived in the U.S. in February of 1986, and there was a Cold Spring Harbor Quantitative Symposium that year where the project was discussed, and PCR was discussed, and a few other things.

100 more paragraphs, each with 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.

  • 2 voices separated automatically, no setup
  • Richard Gibbs — 95%, Interviewer — 5% of the time
  • Export to TXT, DOCX, SRT, and VTT
  • 0:11Early Life and Education1 min
  • 0:56Moving to the U.S. and Initial Involvement1 min
  • 1:48Early Reactions: PCR and the Human Genome Project2 min
  • 3:49Molecular Biology Advances and PCR-driven Techniques2 min
  • 5:46Sequencing the HPRT Locus and Early Automation5 min
  • 10:38Competing Approaches to Genomics and Cloning Technologies2 min
  • 12:53Cost, Quality, and Scalability of Genome Sequencing2 min
  • 14:26Founding the Baylor Sequencing Lab and Early Projects3 min
  • 17:50Collaboration with Other Sequencing Projects1 min
  • 18:27Responses to Celera's Approach and EST Strategies1 min
  • 19:34Data Quality Debates and the Bermuda Meetings1 min
  • 20:14Computational Needs and Phil Green's Contributions1 min
  • 21:07Creating the Baylor Human Genome Sequencing Center1 min
  • 22:07Origins of Celera and Its Influence on the Human Genome Project1 min

4 more chapters, each with its own timestamp and length

The whole 30 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.

  • 18 chapters across 30 minutes
  • Each chapter links to its place in the recording

Richard Gibbs recounts his arrival from Australia to Houston, his work with HPRT and early PCR-driven methods, the technical and social debates around the Human Genome Project (HGP), the development of sequencing technologies and centers (including Baylor's Human Genome Sequencing Center)…

  • PCR and related molecular techniques were the immediate catalysts that enabled practical gene-level sequencing and disease-gene discovery in the late 1980s.
  • Early sequencing required iterative technical advances (automation, cloning strategies, paired-end reads) before whole-genome efforts became feasible.
  • Social and fiscal opposition to the HGP existed but did not, in Gibbs' view, present insurmountable arguments against the project.

Decisions the AI pulled out

  • Gibbs informally declared his lab to be the Human Genome Sequencing Center at Baylor, establishing the center and pursuing external funding.01
  • The community accepted the utility of draft genome sequences as intermediate products in shotgun-based projects.02
  • Use of paired-end sequencing (sequencing both ends of subclones) was adopted to improve shotgun assembly anchoring.03

No need to replay 30 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.

  • 3 decisions pulled from the conversation
  • 8 key points in the full summary
One-page digest

Gibbs' move to Houston in 1986 to work on HPRT under Thomas Caskey introduced him to early human gene cloning and sequencing efforts.

from the recording
Article draft

“Oral History: Richard Gibbs on Early Human Genome Sequencing”: human genome project, sequencing technology, PCR era

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

“Extract a timeline of Gibbs' sequencing milestones (1986–1995)”: 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
  • HPRT
  • PCR
  • Shotgun sequencing
  • Paired-end sequencing
  • BAC/YAC
  • EST
  • Capillary sequencing
  • Draft genome

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
  • 8 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: “NHGRI's Oral History Collection: Interview with Richard Gibbs”, National Human Genome Research Institute, YouTube, license CC BY 4.0. Transcript, chapters, summary and key points by QuillHub.

How it works

How to transcribe an interview

Three steps in your browser, nothing to install. The typing is done for you, so your time goes into checking quotes.

  1. 1

    Upload the interview recording

    Drag in the file from your phone, voice recorder, or Zoom: audio MP3, M4A, WAV, OGG or video MP4, MOV, WEBM. If the interview is on Google Drive or YouTube, paste the link. One file can run up to 10 hours and 5 GB. Then press Start transcription.

  2. 2

    Let AI transcribe it

    QuillHub recognizes the speech, adds punctuation, and splits the text into the interviewer's and the participant's turns with timestamps. Then it breaks the conversation into chapters and writes a summary. A one-hour interview usually takes a few minutes.

  3. 3

    Name the speakers and download

    Replace Speaker A and Speaker B with names or roles, like Interviewer and Participant 3. Check names, dates, and quotes against the audio, then download DOCX or TXT for Word, Google Docs, or NVivo.

What you get

What you get from an interview recording

One upload gives you the text for quotes and a map of the conversation. Everything is tied to the time in the recording, so any answer is easy to find and replay.

Questions and answers by speaker

Each turn starts with the speaker, so the transcript reads as a dialogue. Rename the labels once, and the names carry into the export.

Timestamps on every line

Each line starts with its time in the recording. Click a line to jump the player there, and slow it down to 0.5× to catch a fast or quiet answer.

Chapters

The interview split into topics, each with a start time. Useful when you need the part about methods from a 90-minute conversation.

Summary and key points

A few sentences on what the interview covered and the main points, one line each. Good for a quick memo after every session. Only the transcript is word for word; the summary is a paraphrase.

DOCX, TXT, and Markdown

Choose what goes into the file: the overview, the text by chapters with speaker names, or continuous text with a timestamp about every 30 seconds. DOCX opens in Word and Google Docs and imports into NVivo.

Questions to the whole archive

Ask the AI chat about one interview or all of them: where participants mention cost, who disagreed with whom. Every answer links to the interview it came from.

AI on top of the transcript

Ask your interviews, not just read them

A transcript is where analysis starts. When the study has twelve interviews, ask all of them at once and open the ones that answer.

Onboarding study · participant 752:14
What made participant 7 almost cancel the trial?

The setup step that asks for a company domain. She says she didn't have admin access and waited two days for IT, and that was the moment she nearly gave up.

Participant 7 · 38:40 — the fragment is lit on the track
Which other participants talk about waiting for IT?

Participants 2, 5, and 11. Two of them describe the same domain step; participant 5 mentions it about single sign-on.

12 interviews in the study
P2 · 14:05 · setup
P11 · 22:30 · domain step
01

Questions across all interviews

Ask the chat about every recording in your archive at once and get answers that link to the source.

02

Summary, tasks, or a draft

One click for a summary, a task list, or a blog post draft from an interview. Or write your own prompt.

03

Access for your AI agents

The MCP server lets Claude and other AI agents search your interviews and read the transcripts directly.

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

Ways to do it

Five ways to transcribe an interview

AI software, a human service, the tools you may already have, or typing it yourself. Here is what each one costs you in money and time.

OptionHow it worksSpeakers and timestampsCost and timeGood for
QuillHubUpload a recording or paste a link; AI transcribes and writes a summary and chaptersSpeakers separated automatically, a timestamp on every line60 free minutes, then minute packs from $2.99 or $9.99 a month; a few minutes per hour of audioResearchers, journalists, and teams with many interviews to search and quote
Human transcription serviceYou upload the file, a transcriptionist types it and checks itSpeakers, timestamps, and marks like [inaudible] on requestPaid per audio minute: Rev charges $1.99, about $119 for a one-hour interview, delivered within 12 hoursStrict verbatim with pauses and fillers, or audio too poor for AI
Microsoft Word TranscribeUpload .mp3, .m4a, .wav, or .mp4 inside Word for the webSpeakers separatedMicrosoft 365 subscription, up to 300 minutes of uploaded audio a monthA few interviews a month if you already pay for Microsoft 365
Google Docs voice typingTypes what the computer's microphone hears; you can't upload a fileNo speakers, no timestampsFree in the browser, but in real time: an hour of interview takes an hourDictating your own notes after the interview
Typing it yourselfPlay a few seconds, pause, type, repeatWhatever you mark by handFree, but several hours for every hour of audioA short interview, or when you want to relive every word

Which interviews people transcribe

Any conversation where the exact words matter later: for a quote, a code, or a decision.

Qualitative research interviews

Semi-structured and in-depth interviews for a thesis or a study. Transcripts come out ready for coding in NVivo or by hand in Word.

Journalism and long reads

Find the quote in the text instead of scrubbing through the audio, then click it and check it against the recording before it goes to print.

User research and customer interviews

Every session gets a summary, and the chat finds patterns across the whole round: the same complaint, the same feature request.

Oral history

Long life-story interviews, like the sample on this page, become searchable text with chapters by topic and period.

Focus groups

Several voices are separated by speaker. Name each label once, and the transcript shows who said what.

Hiring interviews

Compare candidates by what they actually said, not by notes taken while trying to listen.

Who it's for

Who uses interview transcription

The more interviews in a project, the more hours a transcript saves.

Researchers and students

Transcripts ready for coding

A semester of interviews no longer means weeks of typing. Upload each recording, check the quotes, and move on to analysis.

  • Interviewer and participant labels you can rename
  • DOCX and TXT for NVivo and Word
  • Find and replace to swap names for pseudonyms

Journalists

The quote in seconds, not an hour of replay

Find the line you remember in the text, click it, and hear it again. Spellings of names get fixed once for the whole interview.

  • Click a line to hear the original
  • Chapters to jump through a long interview
  • Ask the chat across all your interviews
QuillHub for media

Product and UX teams

Customer interviews you can search

Each interview gets its own summary and key points, and the chat answers questions across all of them with links to the interviews it drew on.

  • Summary per interview
  • Answers that link to the source interview
  • Patterns across the whole study
QuillHub for product teams

Recruiters and HR

Every candidate on the record

Listen to the candidate instead of typing. After the interview, the transcript and a summary are ready to share with the hiring manager.

  • Summary of each interview
  • Share link that stays off until you turn it on
  • Compare answers across candidates
QuillHub for HR

Tips

How to get a better interview transcript

The quality of the transcript is decided when you record. A few habits make the text cleaner and the review shorter.

  • Put the microphone close to the participant

    A phone on the table between you works better than a laptop across the room. Turn off fans and step away from cafe music.

  • Ask for consent on the recording

    Ask permission at the start, with the recorder already running. Your ethics board or publication may need it, and the participant often says their name, which helps you label the speakers.

  • Fix misheard names once

    Names and terms come out wrong most often: in the sample above, Wilhelm Ansorge became “Wilhelm and Sorge”. Open Edit text and use Find and replace, and the fix applies to the text, the chapters, and the summary. Tick Save to vocabulary, and new recordings will spell it your way too.

  • Use roles or pseudonyms

    Rename the speakers to Interviewer and Participant 4, and replace the participant's name in the text with a pseudonym. The audio is unchanged, and the original text can be brought back with Restore original text, so keep the recording private.

  • Add the header and marks in Word

    Date, place, participants, and marks like [inaudible] or [laughter] are added by hand. QuillHub gives you the dialogue with speakers and timestamps to build on.

  • Set the language for strong accents

    The language is detected automatically, one per recording. If the participant has a strong accent, pick the language under Audio Language before you press Start transcription.

Without the website

Interview transcription in Telegram

If you use Telegram, you can also send an interview recording 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 recording straight 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

Interview transcription: frequently asked questions

If your question isn't here, contact support.

It is software that turns an interview recording into text: the questions and answers, split by speaker, with timestamps. QuillHub also writes a summary and chapters, so you can find any part of the conversation without replaying it.

A speech recognition model turns the recording into text and tells the voices apart. QuillHub adds punctuation, speaker labels, and timestamps, and then a language model splits the conversation into chapters and writes the summary and key points. Recordings shorter than three minutes get only the text.

Yes. QuillHub gives you 60 free minutes when you sign up, no card required. That covers a one-hour interview with speakers, timestamps, and the summary. After that, minute packs start at $2.99, and Unlimited is $9.99 a month.

Start with a header: the date, the place or platform, and who took part. Write the conversation turn by turn, each turn starting with the speaker's name or role, and add timestamps at each turn or every few minutes. Mark unclear words as [inaudible] with the time. QuillHub gives you the turns, speakers, and timestamps; the header and marks you add in Word.

It is close. The text keeps repetitions, false starts, and words like “you know” and “so”. Fillers like “um” and “uh” are left out, and pauses and laughter are not marked. If your method needs strict verbatim, add those marks by hand while you listen. The summary and chat answers are paraphrases; only the transcript is word for word.

Yes. Recognize Speakers is on by default, and you don't need to say how many people are talking. Each voice gets a label, Speaker A, Speaker B, and so on. Click a label to rename it, and the new name goes into the downloaded file.

Yes. Download DOCX or TXT: NVivo imports both as documents, and so do most coding tools. Choose the text by chapters with speaker names for coding, or the text with timestamps if you want to cite the time of each quote. Coding itself happens in your analysis tool, not in QuillHub.

Yes, voices are separated for groups too, and you don't set the number of speakers in advance. People who talk over each other or sit far from the microphone are harder to tell apart, so check the labels in the busiest parts of the discussion. A microphone in the middle of the table helps the most.

It depends on the audio. The 30-minute oral history on this page reads cleanly, but a few names came out wrong, and “Celera” turned into “CRISPR” once. Cross-talk, noise, and strong accents add errors, and we don't promise an accuracy percentage. Check names, numbers, and every quote you publish against the audio: click the line, and the player jumps there.

Follow the guide your department, university, or journal gives you, since requirements differ. A common layout is a header with the date and participants, the participant under a pseudonym or code, each turn labeled with the speaker, and line numbers or timestamps for citing. QuillHub's DOCX can include the text by chapters with speaker names and chapter start times, and the text with a timestamp about every 30 seconds.

Record the call with your phone's recorder or a call recording app, with the other person's consent, and upload the file. QuillHub doesn't dial in or record phone calls itself. For video calls on your computer, QuillHub Desktop records the call on Mac and Windows without a bot.

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. If your ethics board has rules about where interview data may be processed, read our privacy policy before you upload.

QuillHub recognizes speech in 98+ languages, one language per recording. For an interview in two languages, pick the main one under Audio Language; parts in the other language may come out with errors.

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.

Transcribe your interview for free

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