danvics/Handy
1
0
Fork 0
Code Issues Pull requests Projects Releases Packages Wiki Activity Actions

feat: pass custom words as Whisper initial_prompt instead of post-correction (#1035)

Browse source 
Whisper supports an `initial_prompt` parameter that biases the model's
vocabulary during transcription. For Whisper models, we now join the
user's custom words list into a comma-separated string and pass it as
the initial_prompt. This guides the model to prefer those spellings
during decoding rather than relying solely on fuzzy post-correction.

For non-Whisper engines (Parakeet, Moonshine, SenseVoice, GigaAM),
the existing apply_custom_words post-correction continues to apply
since those engines don't support prompt-based vocabulary biasing.

Co-authored-by: Claude Opus 4.6 <noreply@anthropic.com>
This commit is contained in:
Javier Campanini 2026-03-13 21:37:46 -04:00 committed by GitHub
parent 6ad588cba0
commit 85a8ed77b5
No known key found for this signature in database
GPG key ID: B5690EEEBB952194
1 changed files with 14 additions and 2 deletions
Show stats Download patch file Download diff file Expand all files Collapse all files

16
src-tauri/src/managers/transcription.rs
View file

@ -501,6 +501,11 @@ impl TranscriptionManager {
let params = WhisperInferenceParams {
language: whisper_language,
translate: settings.translate_to_english,
initial_prompt: if settings.custom_words.is_empty() {
None
} else {
Some(settings.custom_words.join(", "))
},
..Default::default()
};
@ -602,8 +607,15 @@ impl TranscriptionManager {
}
};
// Apply word correction if custom words are configured
let corrected_result = if !settings.custom_words.is_empty() {
// Apply word correction if custom words are configured.
// Skip for Whisper models since custom words are already passed as initial_prompt.
let is_whisper = self
.model_manager
.get_model_info(&settings.selected_model)
.map(|info| matches!(info.engine_type, EngineType::Whisper))
.unwrap_or(false);
let corrected_result = if !settings.custom_words.is_empty() && !is_whisper {
apply_custom_words(
&result.text,
&settings.custom_words,