Skip to content

Fix Whisper ASR timestamp regression #1742

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
anshusaurav wants to merge 1 commit into
base: main
Choose a base branch
from
+4 −2
Open

Fix Whisper ASR timestamp regression #1742

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
1 change: 0 additions & 1 deletion packages/transformers/src/generation/logits_process.js
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -304,7 +304,6 @@ export class WhisperTimeStampLogitsProcessor extends LogitsProcessor {

if (input_ids[i].length === this.begin_index) {
batch_logits_data.subarray(0, this.timestamp_begin).fill(-Infinity);
continue;
}

// timestamps have to appear in pairs, except directly before eos_token; mask logits accordingly
Expand Down
5 changes: 4 additions & 1 deletion packages/transformers/src/models/whisper/modeling_whisper.js
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -201,7 +201,9 @@ export class WhisperForConditionalGeneration extends WhisperPreTrainedModel {

// input_features shape: [batch=1, n_mels, total_frames]
const input_features = inputs;
const total_frames = input_features.dims[2];
const total_frames = generation_config.num_frames
? Math.min(input_features.dims[2], generation_config.num_frames)
: input_features.dims[2];

// The encoder downsamples by input_stride (=2 for whisper), so:
// num_segment_frames = input_stride * max_source_positions = 3000 mel frames per segment
Expand Down Expand Up @@ -343,6 +345,7 @@ export class WhisperForConditionalGeneration extends WhisperPreTrainedModel {
if (seek_token_timestamps) {
allTokenTimestamps.push(...seek_token_timestamps.slice(0, tokens_to_keep));
}
if (segment_offset <= 0) break;
seek += segment_offset;
}

Expand Down