Core event logging
Records the writing event stream and produces the standard TypeFlow workbook analyses, including pausing, revision and burst data.
Aitor Garcés-Manzanera (University of Murcia) is the creator and lead developer of TypeFlow.
Specialized application for writing-process researchTypeFlow Speak · Think-aloud research
Keystroke logging with synchronized think-aloud audio
Capture the evolving text, writing events and spoken verbalizations in one browser-based session—then export the original evidence and an aligned automatic transcript for analysis.
No installation · microphone required · research files downloaded locally
Automatic multilingual transcription
Powered by Gemini services
Evidence architecture
One synchronized session
Product, process and verbal-protocol evidence remain identifiable in the export.
01
Writing process
Keystrokes, pauses, revisions, bursts and focus events.
02
Original audio
The participant's recorded verbalizations are the primary speech evidence.
03
Transcript
Time-aligned automatic text for navigation, coding and verification.
Research features
Built on the current TypeFlow Core
Speak retains the TypeFlow writing-event model while adding an audio and transcription layer designed for think-aloud protocols.
Original audio capture
Records the full think-aloud audio alongside the writing task. The downloaded recording remains the primary source for checking verbal data.
Multilingual transcription
Automatically detects spoken languages and natural code-switching. The selected speech profile is preserved as research metadata rather than used as a hard language restriction.
Integrated workbook
Adds dedicated Speak Transcript and Speak Audio Meta sheets to the familiar TypeFlow XLSX structure for traceable quality control.
Research metadata
Stores anonymous participant and session codes, study condition, writing language and expected speech profile across the evidence package.
Resilient evidence capture
If live transcription is interrupted, the original recording continues and the transcript gap is documented for the researcher.
Study workflow
From setup to export
A short, participant-facing workflow keeps configuration separate from the writing task.
STEP 01
Configure the session
Enter anonymous identifiers, writing language and the expected speech profile. Review the recording and transcription consent before continuing.
STEP 02
Write and think aloud
Grant microphone access, start the session and verbalize naturally while writing. The interface reports audio and transcription status throughout.
STEP 03
Stop and retain the files
Allow the export to finish and keep every downloaded file together. Do not close or reload the page while the final evidence package is being prepared.
Research outputs
A complete, auditable evidence package
Every format has a distinct role. Keep the complete set and use the original audio whenever transcript accuracy matters.
TypeFlow workbook
Core sheets plus Speak Transcript, Speak Audio Meta and connection diagnostics.
Original recording
Browser-recorded think-aloud audio; the exact container format depends on browser support.
Final writing product
The participant's final text in an editable document.
Plain-text product
A portable UTF-8 version for analysis pipelines.
Readable transcript
A timestamped document for review and qualitative coding.
Plain-text transcript
A timestamped text file for downstream processing.
Files are downloaded locally. If a researcher address and backup endpoint are configured, TypeFlow also sends an optional server backup request; local downloads remain the authoritative handoff.
Use responsibly
Designed for research and reflective practice
Think-aloud protocols require a clear task design, participant briefing and a documented verification procedure.
Writing-process research
Combine temporal writing data with verbal reports to examine planning, monitoring, problem solving, pausing and revision behaviour.
- Pilot the task, microphone and network conditions
- Preserve anonymous identifiers across every output
- Report how transcript verification was conducted
Teaching and reflection
Use replayable process evidence and think-aloud data to support guided reflection on writing strategies without treating automated output as an assessment score.
- Explain the purpose of recording in advance
- Collect appropriate informed consent
- Apply institutional retention and access rules
FAQs
Before running a study
Do participants need to install software?
No. TypeFlow Speak runs in the browser. A modern desktop browser, a working microphone and permission to download multiple files are required.
Can participants switch between languages while speaking?
Yes. Automatic language detection is designed for multilingual speech and natural code-switching. The expected speech profile selected at setup is stored as metadata and does not lock the session to one language.
Is the automatic transcript a verbatim record?
No automatic transcript should be treated as perfectly verbatim. Use the timestamps to locate speech, then check consequential wording, quotations and coding decisions against the original audio.
What happens if live transcription disconnects?
TypeFlow attempts to reconnect while keeping the original audio recording active. Any resulting transcript gap and connection diagnostics are retained in the exported evidence.
Does it require an internet connection?
The writing logger and browser audio capture run locally, but automatic transcription and the optional backup request require network access. Always run a pilot under the same conditions planned for data collection.
How should audio data be handled?
Microphone audio is processed by a remote transcription service. Researchers should obtain appropriate informed consent and follow their institution's data-protection, access and retention requirements.
Is TypeFlow Speak free to use?
Yes. TypeFlow Speak is provided free of charge for research and teaching, and the participant interface does not request a paid transcription account. Service availability remains subject to network and platform conditions.
Ready to pilot?
Open TypeFlow Speak in a desktop browser
Test microphone access, multilingual speech and the complete export before collecting participant data.