|
|
@ -66,6 +66,11 @@ int main(int argc, char* argv[]) {
|
|
|
|
new ppspeech::CMVN(FLAGS_cmvn_file, std::move(linear_spectrogram)));
|
|
|
|
new ppspeech::CMVN(FLAGS_cmvn_file, std::move(linear_spectrogram)));
|
|
|
|
|
|
|
|
|
|
|
|
ppspeech::FeatureCacheOptions feat_cache_opts;
|
|
|
|
ppspeech::FeatureCacheOptions feat_cache_opts;
|
|
|
|
|
|
|
|
// the feature cache output feature chunk by chunk.
|
|
|
|
|
|
|
|
// frame_chunk_size : num frame of a chunk.
|
|
|
|
|
|
|
|
// frame_chunk_stride: chunk sliding window stride.
|
|
|
|
|
|
|
|
feat_cache_opts.frame_chunk_stride = 1;
|
|
|
|
|
|
|
|
feat_cache_opts.frame_chunk_size = 1;
|
|
|
|
ppspeech::FeatureCache feature_cache(feat_cache_opts, std::move(cmvn));
|
|
|
|
ppspeech::FeatureCache feature_cache(feat_cache_opts, std::move(cmvn));
|
|
|
|
LOG(INFO) << "feat dim: " << feature_cache.Dim();
|
|
|
|
LOG(INFO) << "feat dim: " << feature_cache.Dim();
|
|
|
|
|
|
|
|
|
|
|
@ -105,12 +110,13 @@ int main(int argc, char* argv[]) {
|
|
|
|
if (cur_chunk_size < chunk_sample_size) {
|
|
|
|
if (cur_chunk_size < chunk_sample_size) {
|
|
|
|
feature_cache.SetFinished();
|
|
|
|
feature_cache.SetFinished();
|
|
|
|
}
|
|
|
|
}
|
|
|
|
feature_cache.Read(&features);
|
|
|
|
bool flag = true;
|
|
|
|
if (features.Dim() == 0) break;
|
|
|
|
do {
|
|
|
|
|
|
|
|
flag = feature_cache.Read(&features);
|
|
|
|
feats.push_back(features);
|
|
|
|
feats.push_back(features);
|
|
|
|
sample_offset += cur_chunk_size;
|
|
|
|
|
|
|
|
feature_rows += features.Dim() / feature_cache.Dim();
|
|
|
|
feature_rows += features.Dim() / feature_cache.Dim();
|
|
|
|
|
|
|
|
} while(flag == true && features.Dim() != 0);
|
|
|
|
|
|
|
|
sample_offset += cur_chunk_size;
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
|
|
int cur_idx = 0;
|
|
|
|
int cur_idx = 0;
|
|
|
|