From 21d1bf73bb9d1904a17cad32479493ac8d85ac6d Mon Sep 17 00:00:00 2001
From: Kinan Martin <kinanmartin0@gmail.com>
Date: Fri, 9 May 2025 10:57:41 +0900
Subject: [PATCH] new version of multi_ja_en prepare.sh script which swaps
 Librispeech for MLS English

---
 egs/multi_ja_en/ASR/prepare.sh | 84 ++++++++++++++++++----------------
 1 file changed, 44 insertions(+), 40 deletions(-)

diff --git a/egs/multi_ja_en/ASR/prepare.sh b/egs/multi_ja_en/ASR/prepare.sh
index 7a6a63418..9d21f54c9 100755
--- a/egs/multi_ja_en/ASR/prepare.sh
+++ b/egs/multi_ja_en/ASR/prepare.sh
@@ -19,6 +19,8 @@ vocab_sizes=(
 # All files generated by this script are saved in "data".
 # You can safely remove "data" and rerun this script to regenerate it.
 mkdir -p data
+mkdir -p data/lang
+mkdir -p data/manifests
 
 log() {
   # This function is from espnet
@@ -31,55 +33,54 @@ log "dl_dir: $dl_dir"
 log "Dataset: musan"
 if [ $stage -le 1 ] && [ $stop_stage -ge 1 ]; then
   log "Stage 1: Soft link fbank of musan"
-  mkdir -p data/fbank
   if [ -e ../../librispeech/ASR/data/fbank/.musan.done ]; then
-    cd data/fbank
-    ln -svf $(realpath ../../../../librispeech/ASR/data/fbank/musan_feats) .
-    ln -svf $(realpath ../../../../librispeech/ASR/data/fbank/musan_cuts.jsonl.gz) .
-    cd ../..
+    cd data/manifests
+    mkdir -p musan
+    cd musan
+    ln -svf $(realpath ../../../../../librispeech/ASR/data/fbank/musan_feats) .
+    ln -svf $(realpath ../../../../../librispeech/ASR/data/fbank/musan_cuts.jsonl.gz) .
+    cd ../../..
   else
     log "Abort! Please run ../../librispeech/ASR/prepare.sh --stage 4 --stop-stage 4"
     exit 1
   fi
 fi
 
-log "Dataset: LibriSpeech"
+log "Dataset: MLS English"
 if [ $stage -le 2 ] && [ $stop_stage -ge 2 ]; then
-  log "Stage 1: Soft link fbank of LibriSpeech"
-  mkdir -p data/fbank
-  if [ -e ../../librispeech/ASR/data/fbank/.librispeech.done ]; then
-    cd data/fbank
-    ln -svf $(realpath ../../../../librispeech/ASR/data/fbank/librispeech_cuts*) .
-    ln -svf $(realpath ../../../../librispeech/ASR/data/fbank/librispeech_feats*) .
-    cd ../..
+  log "Stage 2: Soft link manifests (including fbank) of MLS English"
+  if [ -e ../../mls_english/ASR/data/manifests/.mls_english-validated.done ]; then
+    cd data/manifests
+    mkdir -p mls_english
+    cd mls_english
+    ln -svf $(realpath ../../../../../mls_english/ASR/data/manifests/mls_eng_cuts*) .
+    ln -svf $(realpath ../../../../../mls_english/ASR/data/manifests/feats*) .
+    cd ../../..
   else
-    log "Abort! Please run ../../librispeech/ASR/prepare.sh --stage 1 --stop-stage 1 and ../../librispeech/ASR/prepare.sh --stage 3 --stop-stage 3"
+    log "Abort! Please run ../../mls_english/ASR/prepare.sh --stage 1 --stop-stage 1"
     exit 1
   fi
 fi
 
 log "Dataset: ReazonSpeech"
 if [ $stage -le 3 ] && [ $stop_stage -ge 3 ]; then
-  log "Stage 2: Soft link fbank of ReazonSpeech"
-  mkdir -p data/fbank
+  log "Stage 3: Soft link fbank of ReazonSpeech"
   if [ -e ../../reazonspeech/ASR/data/manifests/.reazonspeech.done ]; then
-    cd data/fbank
-    ln -svf $(realpath ../../../../reazonspeech/ASR/data/manifests/reazonspeech_cuts*) .
-    cd ..
-    mkdir -p manifests
-    cd manifests
-    ln -svf $(realpath ../../../../reazonspeech/ASR/data/manifests/feats_*) .
-    cd ../..
+    cd data/manifests
+    mkdir -p reazonspeech
+    cd reazonspeech
+    ln -svf $(realpath ../../../../../reazonspeech/ASR/data/manifests/reazonspeech_cuts*) .
+    ln -svf $(realpath ../../../../../reazonspeech/ASR/data/manifests/feats*) .
+    cd ../../..
   else
     log "Abort! Please run ../../reazonspeech/ASR/prepare.sh --stage 0 --stop-stage 2"
     exit 1
   fi
 fi
 
-# New Stage 3: Prepare char based lang for ReazonSpeech
-if [ $stage -le 3 ] && [ $stop_stage -ge 3 ]; then
+if [ $stage -le 4 ] && [ $stop_stage -ge 4 ]; then
   lang_char_dir=data/lang_char
-  log "Stage 3: Prepare char based lang for ReazonSpeech"
+  log "Stage 4: Prepare char-based lang for ReazonSpeech"
   mkdir -p $lang_char_dir
 
   # Prepare text
@@ -89,7 +90,7 @@ if [ $stage -le 3 ] && [ $stop_stage -ge 3 ]; then
       | ./local/text2token.py -t "char" > $lang_char_dir/text
   fi
 
-  # jp word segmentation for text
+  # Japanese word segmentation
   if [ ! -f $lang_char_dir/text_words_segmentation ]; then
     python3 ./local/text2segments.py \
       --input-file $lang_char_dir/text \
@@ -106,29 +107,31 @@ if [ $stage -le 3 ] && [ $stop_stage -ge 3 ]; then
   fi
 
   if [ ! -f $lang_char_dir/L_disambig.pt ]; then
-    python3 ./local/prepare_char.py --lang-dir data/lang_char
+    python3 ./local/prepare_char.py --lang-dir $lang_char_dir
   fi
 fi
 
-if [ $stage -le 4 ] && [ $stop_stage -ge 4 ]; then
-  log "Stage 4: Prepare Byte BPE based lang"
-  mkdir -p data/fbank
+
+
+if [ $stage -le 5 ] && [ $stop_stage -ge 5 ]; then
+  log "Stage 5: Prepare Byte BPE based lang"
+
+  # Check if required char-based lang data exists
   if [ ! -d ../../reazonspeech/ASR/data/lang_char ] && [ ! -d ./data/lang_char ]; then
     log "Abort! Please run ../../reazonspeech/ASR/prepare.sh --stage 3 --stop-stage 3"
     exit 1
   fi
 
-  if [ ! -d ../../librispeech/ASR/data/lang_bpe_500 ] && [ ! -d ./data/lang_bpe_500 ]; then
-    log "Abort! Please run ../../librispeech/ASR/prepare.sh --stage 5 --stop-stage 5"
+  # Check if BPE data from MLS English exists
+  if [ ! -d ../../mls_english/ASR/data/lang ] ; then
+    log "Abort! Please run ../../mls_english/ASR/prepare.sh --stage 3 --stop-stage 3"
     exit 1
   fi
 
   cd data/
-  # if [ ! -d ./lang_char ]; then
-  #   ln -svf $(realpath ../../../reazonspeech/ASR/data/lang_char) .
-  # fi
-  if [ ! -d ./lang_bpe_500 ]; then
-    ln -svf $(realpath ../../../librispeech/ASR/data/lang_bpe_500) .
+  # Symlink the full lang directory from MLS English (which includes bpe_2000/ and transcript.txt)
+  if [ ! -d ./lang_bpe_2000 ]; then
+    ln -svf $(realpath ../../../mls_english/ASR/data/lang) lang_bpe_2000
   fi
   cd ../
 
@@ -136,7 +139,8 @@ if [ $stage -le 4 ] && [ $stop_stage -ge 4 ]; then
     lang_dir=data/lang_bbpe_${vocab_size}
     mkdir -p $lang_dir
 
-    cat data/lang_char/text data/lang_bpe_500/transcript_words.txt \
+    # Combine Japanese char-level text and English BPE transcript
+    cat data/lang_char/text data/lang_bpe_2000/transcript.txt \
       > $lang_dir/text
 
     if [ ! -f $lang_dir/transcript_chars.txt ]; then
@@ -150,7 +154,7 @@ if [ $stage -le 4 ] && [ $stop_stage -ge 4 ]; then
         --input-file ./data/lang_char/text \
         --output-file $lang_dir/text_words_segmentation
 
-      cat ./data/lang_bpe_500/transcript_words.txt \
+      cat ./data/lang_bpe_2000/transcript.txt \
         >> $lang_dir/text_words_segmentation
     fi