updated bash scripts to incorporate with the speed-perturb arg

egs/aidatatang_200zh/ASR/prepare.sh

@@ -77,7 +77,7 @@ if [ $stage -le 4 ] && [ $stop_stage -ge 4 ]; then
   log "Stage 4: Compute fbank for aidatatang_200zh"
   if [ ! -f data/fbank/.aidatatang_200zh.done ]; then
     mkdir -p data/fbank
-    ./local/compute_fbank_aidatatang_200zh.py
+    ./local/compute_fbank_aidatatang_200zh.py --speed-perturb True
     touch data/fbank/.aidatatang_200zh.done
   fi
 fi

egs/aishell/ASR/local/compute_fbank_aidatatang_200zh.py

@@ -42,7 +42,7 @@ torch.set_num_threads(1)
 torch.set_num_interop_threads(1)
 
 
-def compute_fbank_aidatatang_200zh(num_mel_bins: int = 80):
+def compute_fbank_aidatatang_200zh(num_mel_bins: int = 80, speed_perturb: bool = False):
     src_dir = Path("data/manifests")
     output_dir = Path("data/fbank")
     num_jobs = min(15, os.cpu_count())
@@ -85,7 +85,7 @@ def compute_fbank_aidatatang_200zh(num_mel_bins: int = 80):
                 recordings=m["recordings"],
                 supervisions=m["supervisions"],
             )
-            if "train" in partition:
+            if "train" in partition and speed_perturb:
                 cut_set = (
                     cut_set + cut_set.perturb_speed(0.9) + cut_set.perturb_speed(1.1)
                 )
@@ -109,7 +109,12 @@ def get_args():
         default=80,
         help="""The number of mel bins for Fbank""",
     )
-
+    parser.add_argument(
+        "--speed-perturb",
+        type=bool,
+        default=False,
+        help="Enable 0.9 and 1.1 speed perturbation for data augmentation. Default: False.",
+    )
     return parser.parse_args()
 
 
@@ -119,4 +124,6 @@ if __name__ == "__main__":
     logging.basicConfig(format=formatter, level=logging.INFO)
 
     args = get_args()
-    compute_fbank_aidatatang_200zh(num_mel_bins=args.num_mel_bins)
+    compute_fbank_aidatatang_200zh(
+        num_mel_bins=args.num_mel_bins, speed_perturb=args.speed_perturb
+    )

egs/aishell/ASR/prepare.sh

@@ -114,7 +114,7 @@ if [ $stage -le 3 ] && [ $stop_stage -ge 3 ]; then
   log "Stage 3: Compute fbank for aishell"
   if [ ! -f data/fbank/.aishell.done ]; then
     mkdir -p data/fbank
-    ./local/compute_fbank_aishell.py
+    ./local/compute_fbank_aishell.py --speed-perturb True
     touch data/fbank/.aishell.done
   fi
 fi

egs/aishell/ASR/prepare_aidatatang_200zh.sh

@@ -53,7 +53,7 @@ if [ $stage -le 2 ] && [ $stop_stage -ge 2 ]; then
   log "Stage 2: Process aidatatang_200zh"
   if [ ! -f data/fbank/.aidatatang_200zh_fbank.done ]; then
     mkdir -p data/fbank
-    ./local/compute_fbank_aidatatang_200zh.py
+    ./local/compute_fbank_aidatatang_200zh.py --speed-perturb True
     touch data/fbank/.aidatatang_200zh_fbank.done
   fi
 fi

egs/aishell2/ASR/prepare.sh

@@ -101,7 +101,7 @@ if [ $stage -le 3 ] && [ $stop_stage -ge 3 ]; then
   log "Stage 3: Compute fbank for aishell2"
   if [ ! -f data/fbank/.aishell2.done ]; then
     mkdir -p data/fbank
-    ./local/compute_fbank_aishell2.py
+    ./local/compute_fbank_aishell2.py --speed-perturb True
     touch data/fbank/.aishell2.done
   fi
 fi

egs/aishell4/ASR/prepare.sh

@@ -107,7 +107,7 @@ if [ $stage -le 5 ] && [ $stop_stage -ge 5 ]; then
   log "Stage 5: Compute fbank for aishell4"
   if [ ! -f data/fbank/.aishell4.done ]; then
     mkdir -p data/fbank
-    ./local/compute_fbank_aishell4.py
+    ./local/compute_fbank_aishell4.py --speed-perturb True
     touch data/fbank/.aishell4.done
   fi
 fi

egs/alimeeting/ASR/prepare.sh

@@ -97,7 +97,7 @@ if [ $stage -le 5 ] && [ $stop_stage -ge 5 ]; then
   log "Stage 5: Compute fbank for alimeeting"
   if [ ! -f data/fbank/.alimeeting.done ]; then
     mkdir -p data/fbank
-    ./local/compute_fbank_alimeeting.py
+    ./local/compute_fbank_alimeeting.py --speed-perturb True
     touch data/fbank/.alimeeting.done
   fi
 fi

egs/alimeeting/ASR_v2/local/compute_fbank_alimeeting.py

@@ -25,6 +25,7 @@ It looks for manifests in the directory data/manifests.
 
 The generated fbank features are saved in data/fbank.
 """
+import argparse
 import logging
 from pathlib import Path
 
@@ -48,7 +49,7 @@ torch.set_num_interop_threads(1)
 torch.multiprocessing.set_sharing_strategy("file_system")
 
 
-def compute_fbank_ami():
+def compute_fbank_ami(speed_perturb: bool = False):
     src_dir = Path("data/manifests")
     output_dir = Path("data/fbank")
 
@@ -84,7 +85,10 @@ def compute_fbank_ami():
         suffix="jsonl.gz",
     )
 
-    def _extract_feats(cuts: CutSet, storage_path: Path, manifest_path: Path) -> None:
+    def _extract_feats(
+        cuts: CutSet, storage_path: Path, manifest_path: Path, speed_perturb: bool
+    ) -> None:
+        if speed_perturb:
             cuts = cuts + cuts.perturb_speed(0.9) + cuts.perturb_speed(1.1)
         _ = cuts.compute_and_store_features_batch(
             extractor=extractor,
@@ -109,6 +113,7 @@ def compute_fbank_ami():
         cuts_ihm,
         output_dir / "feats_train_ihm",
         src_dir / "cuts_train_ihm.jsonl.gz",
+        speed_perturb,
     )
 
     logging.info("Processing train split IHM + reverberated IHM")
@@ -117,6 +122,7 @@ def compute_fbank_ami():
         cuts_ihm_rvb,
         output_dir / "feats_train_ihm_rvb",
         src_dir / "cuts_train_ihm_rvb.jsonl.gz",
+        speed_perturb,
     )
 
     logging.info("Processing train split SDM")
@@ -129,6 +135,7 @@ def compute_fbank_ami():
         cuts_sdm,
         output_dir / "feats_train_sdm",
         src_dir / "cuts_train_sdm.jsonl.gz",
+        speed_perturb,
     )
 
     logging.info("Processing train split GSS")
@@ -141,6 +148,7 @@ def compute_fbank_ami():
         cuts_gss,
         output_dir / "feats_train_gss",
         src_dir / "cuts_train_gss.jsonl.gz",
+        speed_perturb,
     )
 
     logging.info("Preparing test cuts: IHM, SDM, GSS (optional)")
@@ -186,8 +194,21 @@ def compute_fbank_ami():
         )
 
 
+def get_args():
+    parser = argparse.ArgumentParser()
+    parser.add_argument(
+        "--speed-perturb",
+        type=bool,
+        default=False,
+        help="Enable 0.9 and 1.1 speed perturbation for data augmentation. Default: False.",
+    )
+    return parser.parse_args()
+
+
 if __name__ == "__main__":
     formatter = "%(asctime)s %(levelname)s [%(filename)s:%(lineno)d] %(message)s"
     logging.basicConfig(format=formatter, level=logging.INFO)
 
-    compute_fbank_ami()
+    args = get_args()
+
+    compute_fbank_ami(speed_perturb=args.speed_perturb)

egs/alimeeting/ASR_v2/prepare.sh

@@ -85,7 +85,7 @@ fi
 if [ $stage -le 5 ] && [ $stop_stage -ge 5 ]; then
   log "Stage 5: Compute fbank for alimeeting"
   mkdir -p data/fbank
-  python local/compute_fbank_alimeeting.py
+  python local/compute_fbank_alimeeting.py --speed-perturb True
   log "Combine features from train splits"
   lhotse combine data/manifests/cuts_train_{ihm,ihm_rvb,sdm,gss}.jsonl.gz - | shuf |\
     gzip -c > data/manifests/cuts_train_all.jsonl.gz

egs/wenetspeech/ASR/prepare.sh

@@ -91,7 +91,7 @@ fi
 if [ $stage -le 3 ] && [ $stop_stage -ge 3 ]; then
   log "Stage 3: Preprocess WenetSpeech manifest"
   if [ ! -f data/fbank/.preprocess_complete ]; then
-    python3 ./local/preprocess_wenetspeech.py
+    python3 ./local/preprocess_wenetspeech.py --speed-perturb True
     touch data/fbank/.preprocess_complete
   fi
 fi