How to Transcribe an Interview:
Fast, Free AI Method

Verbatim, Clean Verbatim or Edited: Choose a Style First

Decide how literal the transcript must be before you start editing. AI transcripts usually land close to clean verbatim: they keep the words but often drop fillers such as “um” and “uh”.

If you need full verbatim, plan a second listening pass to add fillers and pause notes.

How to Transcribe an Interview

Transcribing a recorded interview used to mean hours of pausing, rewinding, and typing. AI transcription compresses that process into three steps:

1. Upload your recording. Go to the Speech to Text tool and drop your interview file onto the upload area. The tool accepts all common audio formats (MP3, WAV, FLAC, OGG, M4A, AAC, WMA) and video formats (MP4, MKV, AVI, MOV, WebM). If you recorded the interview on your phone, the file is typically M4A or MP3 — upload it directly without converting.
2. Choose your settings. Select the output format: TXT for a plain text transcript you can paste into a document, SRT for timestamped subtitles, or VTT for web-compatible captions. For interviews, TXT is usually the best choice. Select the spoken language if you know it, or leave the menu on Auto-detect.
3. Download and edit. Once processing finishes, download the transcript file. Open it in any text editor, Word, or Google Docs. Add speaker labels (e.g., "Interviewer:" and "Respondent:"), fix any misrecognized words, and format the text for your needs — whether that is a journalistic quote sheet, a research coding document, or meeting minutes.

Recording Tips for Better Transcription

The quality of your transcript depends heavily on the quality of your recording. Clean audio means fewer corrections later. Here is how to get it:

• Use an external microphone. Your phone's built-in mic is designed for phone calls at close range, not for capturing a conversation across a table. A lavalier mic clipped near the speakers, or a small tabletop condenser mic placed between participants, dramatically improves voice clarity and reduces ambient noise. Even a basic wired earphone mic placed on the table outperforms a phone sitting two feet away.
• Record in a quiet room. Background noise is the number one killer of transcription accuracy. Coffee shops, open offices, and outdoor locations introduce competing audio that confuses the speech recognition model. Close windows, turn off fans and air conditioning if possible, and avoid rooms with hard surfaces that create echo. A carpeted room with soft furniture absorbs sound reflections and produces cleaner audio.
• Ask speakers not to talk over each other. Overlapping speech is extremely difficult for any transcription system — AI or human — to parse accurately. At the start of the interview, briefly mention that you are recording and ask participants to let each person finish before responding. This small request saves significant editing time later.
• Pick a clear format, then check the size. WAV or FLAC avoids compression artifacts, but one hour of 16-bit, 44.1 kHz stereo WAV is about 635 MB, far over the 100 MB upload limit. MP3 or M4A at 128–192 kbps keeps speech clear: one hour at 128 kbps is 57.6 MB. The tool converts every upload to 16 kHz mono before transcribing, so a higher bitrate adds little. Heavily compressed audio (64 kbps MP3 or lower) strips out subtle consonant sounds and sibilants that the model needs to distinguish between similar words.
• Keep the mic close to speakers. By the inverse square law, doubling the distance between the microphone and the speaker cuts the sound intensity to a quarter, about 6 dB quieter. A mic 6 inches from the speaker captures clear, intelligible audio. The same mic 4 feet away picks up mostly room ambience with speech buried underneath. If you cannot use lapel mics, place the recording device in the center of the group, not at the edge of the table.
• Do a test recording first. Record 30 seconds and play it back before starting the actual interview. Listen for echo, hum, buzzing, or low volume levels. It is much easier to fix problems before the interview than to deal with a degraded transcript afterward. If a finished recording has hum or hiss, our AI noise reduction tool can clean it up before you transcribe.

Interview Transcription for Different Fields

Different professions use interview transcripts in different ways, and each has specific requirements beyond a basic word-for-word text dump.

Journalism

Journalists need accurate direct quotes they can attribute to sources. A single misquoted word can change the meaning of a statement and damage credibility. After AI transcription, journalists should listen back to any passage they plan to quote directly, verifying exact wording against the audio. The AI transcript serves as a searchable index of the conversation — use Ctrl+F to find the section where a specific topic was discussed, then verify the exact quote by ear. For longer investigative pieces, timestamped SRT format can help you jump to the right moment in the recording.

Academic Research

Qualitative researchers conducting semi-structured or unstructured interviews need transcripts for thematic coding and discourse analysis. Academic transcription standards often require noting pauses, laughter, emphasis, and non-verbal cues — details that AI does not capture. Use the AI transcript as your base layer, then do a single pass through the audio to add annotations your methodology requires. For large interview studies (20+ interviews), AI transcription can reduce your total transcription time from weeks to days, freeing you to spend more time on analysis rather than typing.

HR and Recruiting

Hiring managers and recruiters transcribe candidate interviews to compare responses across applicants, share with colleagues who were not present, and maintain records for compliance purposes. AI transcription provides a fast, consistent record of each conversation. Label each speaker (Interviewer / Candidate) and organize the transcript by question for easy side-by-side comparison. Some organizations retain interview transcripts as documentation of their hiring process for equal opportunity compliance.

Legal

Depositions, witness statements, and client consultations often need to be transcribed. Legal transcription demands extremely high accuracy because transcripts may become evidence or part of the court record. AI transcription can produce a useful first draft, but for any document that will be filed with a court or used in proceedings, the transcript must be reviewed word by word against the audio. For informal internal notes (case strategy discussions, client intake calls), AI accuracy is typically sufficient without exhaustive review.

UX Research

User experience researchers conduct usability tests and user interviews to understand how people interact with products. Transcripts feed into affinity diagrams, journey maps, and insight reports. AI transcription excels here because UX interviews are typically conducted in quiet settings with good microphones, and the researcher needs a searchable text for pattern-finding across multiple sessions. Tag each transcript with the participant identifier and session date, then use text search to find recurring themes across all interviews.

Editing Your Transcript

The raw AI transcript is a starting point, not a finished document. Here is a practical workflow for turning it into something usable:

1. Download the TXT file. The plain text format works with every text editor and word processor. Open it in Microsoft Word, Google Docs, LibreOffice, or any editor you prefer.
2. Add speaker labels. The AI outputs a continuous stream of text without identifying who said what. Go through the transcript and insert speaker labels at each change of speaker. For a two-person interview, this is straightforward — you know when you asked a question and when the subject answered. For group interviews or panel discussions, you may need to listen to short segments to identify voices.
3. Clean up recognition errors. AI handles common words well but may stumble on proper nouns (names of people, companies, products), technical jargon, acronyms, and words spoken with heavy accents. Scan through the transcript and correct these. A useful technique: search for common AI misrecognitions in your field and fix them in batch using find-and-replace.
4. Format for publication or analysis. Depending on your purpose, you may need to add paragraph breaks at topic changes, insert timestamps at key moments, bold important quotes, or structure the document with headings. For academic coding, some researchers format transcripts in a two-column table: the left column for the transcript text and the right column for codes and annotations.

AI vs Human Transcription

AI transcription and professional human transcription each have strengths. Choosing the right one depends on your accuracy requirements, budget, and turnaround time.

When AI transcription is enough

• Internal notes and meeting summaries. If the transcript is for your own reference or internal team use, minor errors are easy to overlook or correct as you read.
• Rough draft for further editing. When you plan to rewrite the content anyway — turning an interview into a blog post, article, or report — the AI transcript gives you the raw material to work from.
• High-volume projects. Transcribing 30 user research interviews or 50 candidate screenings is impractical with human transcriptionists on a tight budget and timeline. AI handles the bulk, and you refine the key sections.
• Quick turnaround needs. Breaking news, same-day reports, or time-sensitive research benefits from a transcript that is available in minutes rather than days.

When you need human transcription

• Legal proceedings. Court transcripts, depositions, and official legal documents require certified accuracy. A misheard word in legal testimony can have serious consequences.
• Medical records. Patient interviews, clinical trial recordings, and medical dictation involve specialized terminology where errors could affect patient care or research validity.
• Heavy accents, dialects, or multilingual interviews. When speakers code-switch between languages, use regional dialects, or have strong accents, human transcriptionists who speak those languages outperform AI significantly.
• Poor audio quality. Recordings made in noisy environments, with distant microphones, or on aging equipment benefit from a human listener who can use context to fill in unclear words.
• Verbatim requirements. When you need every "um," "uh," false start, and overlapping utterance captured exactly as spoken — common in linguistic research and some legal contexts — human transcription is more reliable.

For many professionals, the best approach is a hybrid workflow: use AI transcription for the initial draft, then invest human review time only on the sections that require absolute precision.

How to Format an Interview Transcript (Example)

Put the basic details at the top, then start a new paragraph at every change of speaker. Add a timestamp at each new question so you can find the audio again.

Header: project or article name, interviewee, interviewer, date, place, recording length and transcript style.

Interviewer [00:00:12]: Thanks for joining. How did the project start?
Respondent [00:00:18]: It began as a side project with two colleagues. We tested it on our own team first.
Interviewer [00:01:05]: What changed when you opened it to customers?