Black Friday offers.

See offers
Pepys

Blog

The Keyboard Bottleneck

A transcript is only useful if it helps you finish the work that made you record in the first place.

By Pepys · Published

Mr Pepys, deadpan: "I recorded it. Finished Now I have to use it." (with "Finished" crossed out)

The short version

The Keyboard Bottleneck is the manual conversion work between having information and using it: replaying an interview to find a quote, typing an explanation you can already say, or postponing a report because it demands screen time. Start by fixing one repeated handoff between speech, text and listening, then measure the finished work rather than the amount captured.

A folder full of recordings is unfinished work

Recording an interview solves the capture problem. It leaves the reporting problem waiting. Someone still has to find the exact passage, check the context and turn it into a sentence another person can use. Add another recording and that queue gets longer.

This is where a lot of voice software stops being useful: the input gets easier while the output stays awkward. You have audio, a summary and perhaps a chat window. The work needs a checked quotation in a document. The distance between those two states is what we call the Keyboard Bottleneck.

Pepys builds transcription, dictation and listening software, so we have a stake in this argument. Our claim is that the next step deserves more attention than the record button. Voice-Native Work is our name for organizing software around those changes of form.

Follow one interview through to the report

Consider a hypothetical research interview. The researcher records with permission and makes a short note afterward: the participant kept returning to the enrollment form. A week later, while writing the findings, the researcher needs the participant's wording and the point in the conversation where it appeared.

The recording alone makes that a listening task. A timestamped transcript makes it a search-and-check task. Search for enrollment, read the surrounding exchange, then play that passage before quoting it. Keep the quote and its timestamp together in the research notes.

Now write the interpretation separately. The researcher can dictate why the form seemed confusing, then edit that explanation against the evidence. Combining the transcript and interpretation into one polished document too early makes it harder to distinguish the participant's words from the researcher's argument.

This example is a proposed workflow, not a customer result or a speed test. It also starts after the institution has approved the tool and the handling of the recording. Our pre-upload checklist covers those decisions; the research software comparison covers product selection.

Speaking faster does not mean finishing faster

Dictation is appealing when you can explain a point but cannot get the first sentence down. Put the cursor in the email, explain the request with Pepys Dictate and review the result. It gives you something to edit.

But input speed is only one part of writing. A draft full of unclear commitments takes time to repair, however quickly it appeared. Compare the time from opening the draft to sending a version you are willing to stand behind. Include corrections, rearrangement and fact checking.

The often-cited speech-versus-typing study tested text entry on mobile devices. It supports a claim about that experiment, not a promise that dictating an original proposal on a Mac makes the whole job three times faster. Our dictation and typing article links the study. For choosing an app, use the dictation buyer's guide.

Listening needs an exit back to the page

Written information has its own waiting queue: saved reports, papers and drafts that require another stretch of screen time. A document reader lets you listen to prose while keeping the source available.

The source matters. A sentence such as costs increased in the second quarter may be easy to follow by ear. A table showing which costs rose needs inspection. Listening should let you return to that page, not force you to guess what a sequence of spoken numbers meant.

For a draft, hearing it aloud is another review method. Repetition and a sentence that runs out of breath can become easier to notice. Use Pepys Reader for the listening pass, then make the edits in the document. These are separate steps, not an automatic integration.

Try a handoff audit before buying another tool

Pick a repeated job from your own week. Write down its source, its destination and what counts as finished. For example: an interview recording becomes a checked quotation in a report; a spoken explanation becomes an email ready to send; a saved paper becomes notes you can refer to later.

Do the next instance your usual way and note where you replay, copy, retype or stop. On another comparable instance, change one handoff. Use a transcript, dictate the rough draft or listen to the prose. Record the time spent on correction as well as capture. This is a suggested self-test, not a Pepys benchmark.

Keep the change if it helps you finish a comparable job with acceptable errors and less effort. Abandon it if reviewing the output costs more than it saves. Different speakers, accents, documents and work environments will produce different results.

Leave the keyboard where it earns its place

A shared office may be the wrong place to dictate a sensitive email. A dense budget deserves a visual check. Someone may prefer reading to listening, or need a transcript because audio is inaccessible. A workable system leaves those choices open.

Our ambition for voice-native work is for more information to remain usable as people's needs change. Speak when explaining is easier. Read when precision needs your eyes. Listen when that helps you get through the material. Keep an editable source and a way back to the evidence.

Choose the format that helps you finish the task already on your desk.

Share this article

References

  1. 1.Ruan et al.: Speech Is 3x Faster than Typing for English and Mandarin Text Entry on Mobile Devices – Stanford HCI Group

Keep Pepys in your Google results

If our transcription guides are useful, add Pepys as a Preferred Source in Google.

Add Pepys as a preferred source on Google
Explore Voice-Native Work with Pepys