Skip to content

fix(transcription): suppress whisper non-speech tokens #1

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Open
maubrowncow wants to merge 1 commit into
base: main
Choose a base branch
from
Open

fix(transcription): suppress whisper non-speech tokens #1

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
1 commit
Select commit
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
2 changes: 1 addition & 1 deletion Cargo.lock
View file
Open in desktop

Some generated files are not rendered by default. Learn more about how customized files appear on GitHub.

2 changes: 2 additions & 0 deletions native-core/macos/Sources/WhisperCppBridge/WhisperCppBridge.swift
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -56,6 +56,8 @@ public final class WhisperCppSpeechModel {
params.n_threads = Int32(ProcessInfo.processInfo.processorCount)
params.no_context = true
params.no_timestamps = true
params.suppress_blank = true
params.suppress_nst = true
params.print_progress = false
params.print_realtime = false
params.print_timestamps = false
Expand Down
4 changes: 4 additions & 0 deletions native-core/shared/prompts/cleanup-default-instructions.md
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -11,3 +11,7 @@ Clean the dictated transcript into polished written text.
3. Add natural punctuation and capitalization. Preserve the speaker's meaning, wording, and tone otherwise.
Input: can you remind me to call Sarah tomorrow
Output: Can you remind me to call Sarah tomorrow?

4. Drop bracketed non-speech annotations describing background sounds, such as [cough], [Music], (applause), [laughter], [breathing], [♪♪♪], or [inaudible]. Leave brackets that the speaker actually meant (UI labels, citations, parentheticals) alone.
Input: I was walking home [cough] when I saw it.
Output: I was walking home when I saw it.
2 changes: 2 additions & 0 deletions native-core/windows/src/model_whisper_cpp.rs
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -87,6 +87,8 @@ impl WhisperCppPipeline {
params.set_translate(false);
params.set_no_context(true);
params.set_no_timestamps(true);
params.set_suppress_blank(true);
params.set_suppress_nst(true);
params.set_print_special(false);
params.set_print_progress(false);
params.set_print_realtime(false);
Expand Down
Loading