Free speaker labels on every transcript

What is speaker diarization?

Speaker diarization figures out who spoke when. Instead of one block of text, you get a transcript split by speaker, which is the difference between something you have to untangle and something you can read. Without it, a 45-minute interview reads as one unbroken paragraph; with it, every line is attributed, so you can skim straight to what one person said.

How automatic speaker identification works

Hushscript separates the voices in a recording and labels each one automatically, with no toggle to find and no extra step. Interviews come back attributed to interviewer and subject; meetings come back attributed to each participant; multi-guest podcasts keep every voice distinct. The same voice keeps the same label from the first minute to the last, even across a long recording, so you’re never left guessing whether “Speaker 2” near the end is the same person as “Speaker 2” at the start.

Rename, tag, and fix speakers

The first label is rarely the last word. Rename “Speaker 1” to a real name and it updates across the whole transcript in one edit, with no hunting down every instance by hand. Tag sections you’ll want to find again later, and if diarization ever splits or merges two speakers incorrectly, find and replace catches the mislabeled lines across the transcript instead of making you fix them one at a time. A 40-minute interview with one speaker mislabeled for two minutes in the middle is a find/replace and a rename, not a re-upload; the editor is built for this kind of quick correction, not a full re-transcription.

Multichannel: when speakers are already split

Some recordings arrive pre-split, like a call recorder that puts each side of the line on its own audio channel. Turn on multichannel and Hushscript uses those channels directly to separate speakers instead of inferring who’s talking from the audio alone, which holds up better when the channels are already isolated. This comes up most with call recordings, some webinar platforms, and dedicated interview recorders that already give each participant a dedicated line. It’s an option, not a requirement: leave it off for a normal single-track recording, and automatic diarization takes over as usual.

Speaker names flow into the rest of the workflow

Speaker labels, renamed or not, carry through merged recordings, the editor, subtitles, documents, and every export format, so the result stays readable well past the first pass. Rename once, at the transcript level, and a chapter breakdown, a pulled quote, or a subtitle file all agree on who’s who without any extra work. When you choose Generate Insights, one optional free run creates the complete result against the current transcript revision: summaries, actions, decisions, open questions, quotes, chapters, speaker topics, subject clusters, export suggestions, terminology and profanity findings, and cleanup. Review cleanup in Edit and apply only the changes you approve.

Why we don’t charge for it

A lot of tools gate speaker separation behind a paid tier or cap how many speakers you can identify. We don’t: it’s free on every transcript, because a transcript without it is only half-useful. What you’re paying for is transcription itself; speaker labels ride along at no extra charge.

Interviews, meetings, and podcasts all read better labeled

Anywhere more than one person talks, speaker labels do the heavy lifting. A one-on-one interview comes back as a clean back-and-forth between interviewer and subject instead of a wall of text you have to attribute yourself. A meeting with five participants reads as five distinct voices, so you can tell who committed to what. A multi-guest podcast episode keeps every guest separate through an hour of crosstalk, which matters when you’re pulling a quote for show notes later. → Podcasts · Audio to text

Private by design

Your audio is deleted once the transcript is ready. Transcripts are encrypted at rest, so a data leak would expose only ciphertext, not your words.