Skip to content

[src] Make ivector-extract-online2.cc ignore pitch in case of dim mismatch #3742

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

Merged
merged 6 commits into from
Feb 3, 2020
Merged

[src] Make ivector-extract-online2.cc ignore pitch in case of dim mismatch #3742

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
6 commits
Select commit Hold shift + click to select a range
3ebe1c3
[src] Making ivector extractor tolerate dim mismatch due to pitch
danpovey Nov 21, 2019
b205347
Fix issue getting expected feature dim
danpovey Nov 22, 2019
a6cd237
Merge remote-tracking branch 'upstream/master' into ivector_dim_fix
danpovey Nov 27, 2019
b63d0ea
[src] Fixes to getting ivector dim to remove pitch.
danpovey Nov 27, 2019
01a6abe
[src] Fix one more issue in removing pitch from ivector-feature extra…
danpovey Nov 28, 2019
ba75d7b
[src] one more fix to ivector-dim-ignore-pitch thing
danpovey Nov 29, 2019
File filter

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
9 changes: 9 additions & 0 deletions src/online2/online-ivector-feature.cc
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -67,6 +67,15 @@ void OnlineIvectorExtractionInfo::Init(
this->Check();
}

int32 OnlineIvectorExtractionInfo::ExpectedFeatureDim() const {
int32 num_splice = 1 + splice_opts.left_context + splice_opts.right_context,
full_dim = lda_mat.NumCols();
if (!(full_dim % num_splice == 0 || full_dim % num_splice == 1)){
KALDI_WARN << "Error getting expected feature dimension: full-dim = "
<< full_dim << ", num-splice = " << num_splice;
}
return full_dim / num_splice;
}

void OnlineIvectorExtractionInfo::Check() const {
KALDI_ASSERT(global_cmvn_stats.NumRows() == 2);
Expand Down
2 changes: 2 additions & 0 deletions src/online2/online-ivector-feature.h
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -193,6 +193,8 @@ struct OnlineIvectorExtractionInfo {

void Init(const OnlineIvectorExtractionConfig &config);

int32 ExpectedFeatureDim() const;

// This constructor creates a version of this object where everything
// is empty or zero.
OnlineIvectorExtractionInfo();
Expand Down
14 changes: 13 additions & 1 deletion src/online2bin/ivector-extract-online2.cc
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -94,6 +94,7 @@ int main(int argc, char *argv[]) {
RandomAccessBaseFloatVectorReader frame_weights_reader(frame_weights_rspecifier);
BaseFloatMatrixWriter ivector_writer(ivectors_wspecifier);

bool warned_dim = false;
for (; !spk2utt_reader.Done(); spk2utt_reader.Next()) {
std::string spk = spk2utt_reader.Key();
const std::vector<std::string> &uttlist = spk2utt_reader.Value();
Expand All @@ -108,7 +109,18 @@ int main(int argc, char *argv[]) {
}
const Matrix<BaseFloat> &feats = feature_reader.Value(utt);

OnlineMatrixFeature matrix_feature(feats);
int32 feat_dim = feats.NumCols();
if (feat_dim == ivector_info.ExpectedFeatureDim() + 3) {
if (!warned_dim) {
KALDI_WARN << "Feature dimension is too large by 3, assuming there are "
"pitch features and removing the last 3 dims.";
warned_dim = true;
}
feat_dim -= 3;
}

SubMatrix<BaseFloat> range = feats.ColRange(0, feat_dim);
OnlineMatrixFeature matrix_feature(range);

OnlineIvectorFeature ivector_feature(ivector_info,
&matrix_feature);
Expand Down