mirror of
https://github.com/k2-fsa/icefall.git
synced 2025-08-08 09:32:20 +00:00
Small fix (#1853)
This commit is contained in:
parent
48088cb807
commit
a2b0f6057c
@ -10,9 +10,9 @@ stop_stage=4
|
||||
|
||||
dl_dir=$PWD/download
|
||||
|
||||
dataset_parts="Premium" # Basic for all 10k hours data, Premium for about 10% of the data
|
||||
dataset_parts="Premium" # Basic for all 7226 hours data, Premium for 945 hours subset.
|
||||
|
||||
text_extractor="pypinyin_initials_finals" # default is espeak for English
|
||||
text_extractor="pypinyin_initials_finals" # default is espeak for English
|
||||
audio_extractor="Encodec" # or Fbank
|
||||
audio_feats_dir=data/tokenized
|
||||
|
||||
@ -63,7 +63,7 @@ if [ $stage -le 2 ] && [ $stop_stage -ge 2 ]; then
|
||||
--audio-extractor ${audio_extractor} \
|
||||
--batch-duration 2500 --prefix "wenetspeech4tts" \
|
||||
--src-dir "data/manifests" \
|
||||
--split 100 \
|
||||
--split 100 \
|
||||
--output-dir "${audio_feats_dir}/wenetspeech4tts_${dataset_parts}_split_100"
|
||||
cp ${audio_feats_dir}/wenetspeech4tts_${dataset_parts}_split_100/unique_text_tokens.k2symbols ${audio_feats_dir}
|
||||
fi
|
||||
|
Loading…
x
Reference in New Issue
Block a user