What transcription actually is

This is Talk Type, from the team at Pepys, where we turn talk into text.

What is transcription, actually? Strip it down, and transcription is turning spoken audio into written text. Someone talks. You end up with words on a page you can read, search, and copy. That's it. But the word hides a few choices that matter more than people expect.

The first choice is who does it. For a long time, transcription meant a person, wearing headphones, typing what they heard, pausing and rewinding for hours. Accurate, but slow, and expensive. Then came speech to text, where software does the listening.

And that's the second thing worth pulling apart, because people use transcription and speech to text like they mean the same thing, and they almost do. Speech to text is the engine. It's the model that takes an audio signal and predicts the words. Transcription is the whole job. The engine, plus everything around it that makes the result actually usable.

That everything around it is where the real work is. A raw wall of text is technically a transcript, but it's miserable to use. So a good transcript adds structure. Punctuation and paragraphs, so it reads like writing instead of one long run on sentence. Speaker labels, so you can tell who said what in an interview or a meeting. Timestamps, so you can jump back to the exact moment in the audio and check a quote. Take those away and you've got a search box full of words with no map.

There are also flavors of transcription, and the difference is about how much you keep. Verbatim keeps everything. Every um, every false start, every stutter. You want that for legal work, or research where the exact words matter. Clean verbatim strips the filler and tidies the grammar, so it reads smoothly. That's what most people actually want for notes, articles, and show notes.

So when someone says just transcribe this, they're usually asking for more than raw words. They want speech to text, plus structure, in the flavor that fits the job. Knowing that difference is the whole game. It's the difference between a file you fight with, and one you can actually use.

That's this episode of Talk Type. The full write up, with the links and sources, is in the show notes. Pepys transcribes any file or link, any length, pay once, and we never train on your audio. Your first sixty minutes are free at pepys dot co. Thanks for listening, and we'll see you next time.