Clinician-controlled transcript review
SALT Transcript Editor
The SALT transcript editor lets clinicians turn an AI draft into a reviewed language sample. It supports familiar speaker lines and transcript codes for mazes, omissions, errors, pauses, unintelligible words, and overlap.
The editor is designed for clinical control. You can accept or reject coding suggestions, inspect a live preview, save the coded transcript, and re-analyze the sample before generating a report.
Sample result
SALT Transcript Editor
Before
C: He runned to the playground.
After
C: He runned [EW] to the playground.
Effect
The report reflects the reviewed grammar code.
Speaker lines
C:, E:, P:
Codes
Mazes, errors, omissions
Review
Diff and suggestion controls
Output
Instant re-analysis
Speaker lines
C:, E:, P:
Codes
Mazes, errors, omissions
Review
Diff and suggestion controls
Output
Instant re-analysis
How it fits into a speech workflow
1
Collect
Start from a recording, transcript, or saved session.
2
Review
Check speaker turns and make clinical edits before relying on results.
3
Measure
See the language measures and notes that matter for this feature.
4
Use
Bring the output into reports, progress review, or research exports.
Edit the transcript before trusting the numbers
Automated transcription is useful, but clinical language sample analysis still needs expert review. The editor gives SLPs a focused place to mark errors, remove maze material from core counts, and confirm the utterance-level transcript used for the analysis.
Codes clinicians recognize
The editor supports common SALT-style conventions rather than inventing a new coding language. Speaker prefixes, parenthesized maze material, bracketed error codes, omitted morphemes, and unintelligible segments are all visible in the editor and preview.
- Mazes such as `(um)`, repetitions, revisions, and abandoned starts.
- Errors such as `[EW]`, `[EU]`, and target forms like `[EO:went]`.
- Omissions such as `*ed`, `*s`, and `*the`.
- Unintelligible material such as `xx` and `xxx`.
Live preview and re-analysis
As the transcript changes, the preview updates the sample lines and key metrics. Saving the transcript sends the reviewed text back through the analysis workflow so the final report is based on the edited version, not the original AI draft.
What users see
Sample coded line
A compact result view turns the feature into reviewable language, not a technical readout.
Before
C: He runned to the playground.
After
C: He runned [EW] to the playground.
Effect
The report reflects the reviewed grammar code.
Clinical interpretation notes
- The editor supports the common codes ConductSpeech parses today; it is not a full clone of every SALT desktop command.
- Clinical review is still required before using the transcript in an official report.
Related pages
Maze Analysis for Language Samples
Review filled pauses, repetitions, revisions, abandoned starts, and maze words per 100 words in language samples.
Clinical Language Sample Reports
Generate IEP-ready language sample reports with MLU, PGU, SI, C-units, maze summaries, norms, and fluency context.
Inter-Rater Reliability for Transcripts
Compare two independent language sample transcripts and review agreement for training and QA workflows.
SALT-compatible analysis methodology
Read how ConductSpeech documents conventions, validation, and limitations.
Ready to try it
Start with a real language sample.
Create an account, upload or review a sample, and see how this feature appears inside the ConductSpeech workflow.