2 credits a minute

Podcast Transcription with Speaker Names and Timestamps

Podcast transcription turns an episode into text with every speaker and every timestamp. Upload an episode of up to 3 hours, rename Speaker 1 to your host, merge two labels that are one person, and download the podcast transcript as VTT for Apple Podcasts, SRT for YouTube captions, and the JSON and HTML that your feed and episode page use. On long episodes, labels split across the recording’s parts are merged back to one per person (9 → 4 on a 59-minute interview).

credits, by the started minute
2 / min
one episode, up to 1 GB
3 hours
labels merged on a 59-minute interview
9 → 4
for Apple Podcasts and YouTube
VTT · SRT
Podcasting 2.0 transcript files
JSON · HTML
covered by the sign-up credits
30 min

How do you get a podcast transcript?

Open podcast transcription and upload your episode. In a few minutes you have the podcast transcript with each speaker and timestamp. Rename the speakers, merge any two that are the same person, then download VTT for Apple Podcasts, SRT for YouTube and the JSON or HTML for your feed. It costs 2 credits a minute; the sign-up credits cover 30 minutes.

Transcribe podcast episodes in three steps

From podcast to text in three steps, on any device.

  1. Upload the episode

    Drop the MP3, M4A, WAV, FLAC or OGG, or a video podcast. An episode can run 3 hours; the button shows the credits before anything starts.

  2. Check the speakers

    Rename each speaker once and the name changes everywhere. Merge two labels for one person, or give a paragraph to whoever really said it.

  3. Download the files

    Download the podcast transcript as VTT with names for Apple Podcasts, SRT without names for YouTube, transcript JSON for your RSS feed, HTML for the episode page, plus DOCX, TXT and CSV.

Podcast transcription that knows who is speaking

Every podcast transcript comes with speakers and times, editable before you download it.

A lighthouse beam sweeping across the sea at dusk

Speakers you can rename and merge

Each paragraph carries its speaker. Rename Speaker 1 to your host and every export follows. When one person has two labels, choose Same person as and the two become one; when a line went to the wrong person, give that paragraph to the right one.

One label per person on long episodes

Episodes over 19 minutes are transcribed in parts, and a speaker who stays quiet while two parts overlap could come back under a new label. Here the audio model hears a few seconds of each new label beside the earlier speakers and merges the ones that match. On a 59-minute interview with three people and a launch recording, 9 labels became the 4 voices it holds, in all 7 runs.

A podcast transcript in VTT for Apple Podcasts

Apple Podcasts takes VTT or SRT transcripts, and a VTT with speaker names shows who is talking. The VTT here marks each change of speaker with a voice tag, and its lines stay under the 65 characters that Podcasting 2.0 recommends.

SRT without names for YouTube

YouTube captions read best as the words alone, so the SRT for YouTube leaves the speaker names out and keeps the same timings, two lines of up to 42 characters each.

Transcript files for your RSS feed

The podcast:transcript tag of Podcasting 2.0 points players to a transcript file. Download the JSON it describes, with a version and segments of speaker, start, end and words, or the VTT, and link it from each episode.

A transcript for the episode page

The HTML transcript is the format Podcasting 2.0 gives for episode pages: each paragraph with its speaker, its start time and the words, ready to paste into your site. DOCX and TXT have the same names and times.

Priced by the minute, episodes up to 3 hours

Podcast transcription costs 2 credits a minute of audio, by the started minute; no subscription needed. An episode can be 3 hours and 1 GB, and the speaker merge on long episodes is part of the price.

Podcast transcription samples

Podcast to text on four episodes: a 59-minute public-domain interview, its speakers merged across the parts, and this site’s own AI episodes transcribed again and checked against their scripts. Each card plays with the word lit and shows the TXT, VTT, JSON and HTML as made.

How to transcribe podcast episodes

Turning a podcast to text works best from the final mix of the episode, the file you publish. Upload it in the studio, keep the language on Detect or pick it, and start the transcription. Speakers are on, and the button shows the credits for the episode’s length first.

When it is done, read the first minutes to check the names: rename each speaker once, merge any two labels that are one person, and fix a guest’s name where it was spelled the way it sounds. Find and replace changes a name everywhere at once. To make an episode from a document instead, the AI podcast generator writes it and voices it with two hosts.

  1. 1Upload the episode in the podcast transcription studio.
  2. 2Rename the speakers and merge any repeated label.
  3. 3Download the files for your platforms.

Podcast transcript formats and what each one is for

VTT is a caption file with times and, here, speaker names in voice tags; Apple Podcasts shows those names. SRT is the same captions without markup, the file YouTube and most video tools take. The transcript JSON lists segments with speaker, start and end in seconds and the words, the format Podcasting 2.0 defines for feeds. HTML is a page of paragraphs with the speaker in cite, the start in time and the words in p.

TXT and DOCX are for reading and editing: a paragraph per turn with the speaker and time above it. CSV puts each paragraph in a row for a spreadsheet, and our own JSON keeps every word with its time.

Giving Apple Podcasts and YouTube your transcript

Apple Podcasts makes its own transcripts for many shows, and also reads yours: a VTT or SRT linked from your RSS feed, or uploaded for subscription shows. A VTT with speaker names lets listeners see who is talking. Apple lists the languages it transcribes and may skip episodes over 10 hours.

For YouTube, add the SRT as captions on the video. For other apps that support Podcasting 2.0, link the JSON or VTT with the podcast:transcript tag in each episode of your feed.

Podcast transcription for long episodes and several speakers

An episode of up to 3 hours goes up as one file. Recordings over 19 minutes are transcribed in 15-minute parts that overlap by a minute and joined at a pause, speakers matched by the words both parts heard. Someone quiet during that minute is then heard by the audio model against the earlier speakers, so a three-person show keeps three names.

Two people talking over each other, a very short line or a guest who sounds like the host can still end up under the wrong label. The merge and the paragraph menu fix those in a few clicks, and every export follows.

How accurate is a podcast transcript?

On this site’s own AI episodes, compared word by word with their scripts, the transcripts had 0.3 to 6.2% of words wrong in English, the highest where chemical formulas were spoken, and 0.4 to 0.5% of characters in Chinese, with every word given to the right host. Clear speech on a good microphone reads best; music under the voice, crosstalk and strong accents add errors, so read the names and numbers before you publish.

Podcast transcription credits and limits

Podcast transcription is 2 credits a minute, counted by the started minute of the episode, and needs an account. The credits you get when you sign up cover 30 minutes. No subscription needed: credit packs and plans are on the pricing page.

What podcasters do with a podcast transcript

Turn a podcast to text once and use it in every place it helps.

Three chairs turned toward each other around a small round table

A long interview, every voice named

Host and guests keep one name each through the whole episode, ready for show notes and quotes.

Two umbrellas side by side at a doorway on a rainy evening

Two hosts, one conversation

A weekly chat between two hosts, each line marked with who said it.

A woman and a man talking across a small cafe table

An interview with a guest

Questions and answers kept apart, so a guest’s best line is easy to find and quote.

A green plant on a sunny windowsill

A solo episode

One voice, paragraphs and times, for an episode page that search engines can read.

Two beehives on a city rooftop at sunrise

A guest who knows their craft

An expert walks through their work, every tip with its time, ready for show notes and captions.

Podcast transcription FAQ

Start your podcast transcription

Turn your podcast to text: upload an episode, name the speakers and download every podcast transcript file your platforms need.

Transcribe a podcast