add manifests for whisper

k2-fsa · Feb 22, 2024 · 910e5db · 910e5db
1 parent be001a8
commit 910e5db
Show file tree

Hide file tree

Showing 5 changed files with 115 additions and 54 deletions.
diff --git a/egs/aishell4/ASR/local/compute_fbank_aishell4.py b/egs/aishell4/ASR/local/compute_fbank_aishell4.py
@@ -29,7 +29,7 @@
 from pathlib import Path
 
 import torch
-from lhotse import ChunkedLilcomHdf5Writer, CutSet, Fbank, FbankConfig
+from lhotse import ChunkedLilcomHdf5Writer, CutSet, WhisperFbank, WhisperFbankConfig, Fbank, FbankConfig
 from lhotse.recipes.utils import read_manifests_if_cached
 
 from icefall.utils import get_executor, str2bool
@@ -42,10 +42,10 @@
 torch.set_num_interop_threads(1)
 
 
-def compute_fbank_aishell4(num_mel_bins: int = 80, perturb_speed: bool = False):
+def compute_fbank_aishell4(num_mel_bins: int = 80, perturb_speed: bool = False, whisper_fbank: bool = False):
     src_dir = Path("data/manifests/aishell4")
     output_dir = Path("data/fbank")
-    num_jobs = min(15, os.cpu_count())
+    num_jobs = min(8, os.cpu_count())
 
     dataset_parts = (
         "train_S",
@@ -70,7 +70,10 @@ def compute_fbank_aishell4(num_mel_bins: int = 80, perturb_speed: bool = False):
         dataset_parts,
     )
 
-    extractor = Fbank(FbankConfig(num_mel_bins=num_mel_bins))
+    if whisper_fbank:
+        extractor = WhisperFbank(WhisperFbankConfig(num_filters=num_mel_bins, device='cuda'))
+    else:
+        extractor = Fbank(FbankConfig(num_mel_bins=num_mel_bins))
 
     with get_executor() as ex:  # Initialize the executor only once.
         for partition, m in manifests.items():
@@ -121,7 +124,12 @@ def get_args():
         default=False,
         help="Enable 0.9 and 1.1 speed perturbation for data augmentation. Default: False.",
     )
-
+    parser.add_argument(
+        "--whisper-fbank",
+        type=str2bool,
+        default=False,
+        help="Use WhisperFbank instead of Fbank. Default: False.",
+    )
     return parser.parse_args()
 
 
@@ -132,5 +140,5 @@ def get_args():
 
     args = get_args()
     compute_fbank_aishell4(
-        num_mel_bins=args.num_mel_bins, perturb_speed=args.perturb_speed
+        num_mel_bins=args.num_mel_bins, perturb_speed=args.perturb_speed, whisper_fbank=args.whisper_fbank
     )
diff --git a/egs/aishell4/ASR/prepare.sh b/egs/aishell4/ASR/prepare.sh
@@ -5,8 +5,8 @@ export PROTOCOL_BUFFERS_PYTHON_IMPLEMENTATION=python
 
 set -eou pipefail
 
-stage=-1
-stop_stage=100
+stage=20
+stop_stage=20
 perturb_speed=true
 
 
@@ -76,14 +76,24 @@ if [ $stage -le 1 ] && [ $stop_stage -ge 1 ]; then
 fi
 
 if [ $stage -le 2 ] && [ $stop_stage -ge 2 ]; then
-  log "Stage 2: Process aishell4"
+  log "Stage 2: Compute fbank for aishell4"
   if [ ! -f data/fbank/aishell4/.fbank.done ]; then
     mkdir -p data/fbank/aishell4
     ./local/compute_fbank_aishell4.py --perturb-speed ${perturb_speed}
     touch data/fbank/aishell4/.fbank.done
   fi
 fi
 
+whisper_mel_bins=80
+if [ $stage -le 20 ] && [ $stop_stage -ge 20 ]; then
+  log "Stage 20: Compute whisper fbank for aishell4"
+  if [ ! -f data/fbank/aishell4/.fbank.done ]; then
+    mkdir -p data/fbank/aishell4
+    ./local/compute_fbank_aishell4.py --perturb-speed ${perturb_speed} --num-mel-bins ${whisper_mel_bins} --whisper-fbank true
+    touch data/fbank/aishell4/.fbank.done
+  fi
+fi
+
 if [ $stage -le 3 ] && [ $stop_stage -ge 3 ]; then
   log "Stage 3: Prepare musan manifest"
   # We assume that you have downloaded the musan corpus
@@ -106,16 +116,7 @@ if [ $stage -le 4 ] && [ $stop_stage -ge 4 ]; then
 fi
 
 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 --perturb-speed ${perturb_speed}
-    touch data/fbank/.aishell4.done
-  fi
-fi
-
-if [ $stage -le 6 ] && [ $stop_stage -ge 6 ]; then
-  log "Stage 6: Prepare char based lang"
+  log "Stage 5: Prepare char based lang"
   lang_char_dir=data/lang_char
   mkdir -p $lang_char_dir
 

diff --git a/egs/alimeeting/ASR/local/compute_fbank_alimeeting.py b/egs/alimeeting/ASR/local/compute_fbank_alimeeting.py
@@ -29,7 +29,7 @@
 from pathlib import Path
 
 import torch
-from lhotse import CutSet, Fbank, FbankConfig, LilcomChunkyWriter
+from lhotse import CutSet, WhisperFbank, WhisperFbankConfig, Fbank, FbankConfig, LilcomChunkyWriter
 from lhotse.recipes.utils import read_manifests_if_cached
 
 from icefall.utils import get_executor, str2bool
@@ -42,10 +42,10 @@
 torch.set_num_interop_threads(1)
 
 
-def compute_fbank_alimeeting(num_mel_bins: int = 80, perturb_speed: bool = False):
+def compute_fbank_alimeeting(num_mel_bins: int = 80, perturb_speed: bool = False, whisper_fbank: bool = False):
     src_dir = Path("data/manifests/alimeeting")
     output_dir = Path("data/fbank")
-    num_jobs = min(15, os.cpu_count())
+    num_jobs = min(8, os.cpu_count())
 
     dataset_parts = (
         "train",
@@ -70,7 +70,10 @@ def compute_fbank_alimeeting(num_mel_bins: int = 80, perturb_speed: bool = False
         dataset_parts,
     )
 
-    extractor = Fbank(FbankConfig(num_mel_bins=num_mel_bins))
+    if whisper_fbank:
+        extractor = WhisperFbank(WhisperFbankConfig(num_filters=num_mel_bins, device='cuda'))
+    else:
+        extractor = Fbank(FbankConfig(num_mel_bins=num_mel_bins))
 
     with get_executor() as ex:  # Initialize the executor only once.
         for partition, m in manifests.items():
@@ -121,7 +124,12 @@ def get_args():
         default=False,
         help="Enable 0.9 and 1.1 speed perturbation for data augmentation. Default: False.",
     )
-
+    parser.add_argument(
+        "--whisper-fbank",
+        type=str2bool,
+        default=False,
+        help="Use the Whisper Fbank feature extractor. Default: False.",
+    )
     return parser.parse_args()
 
 
@@ -132,5 +140,5 @@ def get_args():
 
     args = get_args()
     compute_fbank_alimeeting(
-        num_mel_bins=args.num_mel_bins, perturb_speed=args.perturb_speed
+        num_mel_bins=args.num_mel_bins, perturb_speed=args.perturb_speed, whisper_fbank=args.whisper_fbank
     )
diff --git a/egs/alimeeting/ASR/prepare.sh b/egs/alimeeting/ASR/prepare.sh
@@ -66,13 +66,22 @@ if [ $stage -le 1 ] && [ $stop_stage -ge 1 ]; then
 fi
 
 if [ $stage -le 2 ] && [ $stop_stage -ge 2 ]; then
-  log "Stage 2: Process alimeeting"
+  log "Stage 2: compute fbank for alimeeting"
   if [ ! -f data/fbank/alimeeting/.fbank.done ]; then
     mkdir -p data/fbank/alimeeting
     ./local/compute_fbank_alimeeting.py --perturb-speed ${perturb_speed}
   fi
 fi
 
+whisper_mel_bins=80
+if [ $stage -le 20 ] && [ $stop_stage -ge 20 ]; then
+  log "Stage 20: compute whisper fbank for alimeeting"
+  if [ ! -f data/fbank/alimeeting/.fbank.done ]; then
+    mkdir -p data/fbank/alimeeting
+    ./local/compute_fbank_alimeeting.py --perturb-speed ${perturb_speed} --num-mel-bins ${whisper_mel_bins} --whisper-fbank true
+  fi
+fi
+
 if [ $stage -le 3 ] && [ $stop_stage -ge 3 ]; then
   log "Stage 3: Prepare musan manifest"
   # We assume that you have downloaded the musan corpus
@@ -95,16 +104,7 @@ if [ $stage -le 4 ] && [ $stop_stage -ge 4 ]; then
 fi
 
 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 --perturb-speed True
-    touch data/fbank/.alimeeting.done
-  fi
-fi
-
-if [ $stage -le 6 ] && [ $stop_stage -ge 6 ]; then
-  log "Stage 6: Prepare char based lang"
+  log "Stage 5: Prepare char based lang"
   lang_char_dir=data/lang_char
   mkdir -p $lang_char_dir
 

diff --git a/egs/multi_zh-hans/ASR/prepare.sh b/egs/multi_zh-hans/ASR/prepare.sh
@@ -60,7 +60,7 @@ if [ $stage -le 2 ] && [ $stop_stage -ge 2 ]; then
 
   if [ ! -f data/fbank/.thchs30.done ]; then
     mkdir -p data/fbank
-    ./local/compute_fbank_thchs30.py
+    ./local/compute_fbank_thchs30.py --speed-perturb true
     touch data/fbank/.thchs30.done
   fi
 fi
@@ -137,7 +137,7 @@ if [ $stage -le 6 ] && [ $stop_stage -ge 6 ]; then
 
   if [ ! -f data/fbank/.stcmds.done ]; then
     mkdir -p data/fbank
-    ./local/compute_fbank_stcmds.py
+    ./local/compute_fbank_stcmds.py --speed-perturb true
     touch data/fbank/.stcmds.done
   fi
 fi
@@ -151,15 +151,15 @@ if [ $stage -le 7 ] && [ $stop_stage -ge 7 ]; then
     lhotse download primewords $dl_dir/primewords
   fi
 
-  if [ ! -f data/manifests/.stcmds.done ]; then
+  if [ ! -f data/manifests/.primewords.done ]; then
     mkdir -p data/manifests
-    lhotse prepare stcmds $dl_dir/primewords data/manifests/primewords
+    lhotse prepare primewords $dl_dir/primewords data/manifests/primewords
     touch data/manifests/.primewords.done
   fi
 
   if [ ! -f data/fbank/.primewords.done ]; then
     mkdir -p data/fbank
-    ./local/compute_fbank_primewords.py
+    ./local/compute_fbank_primewords.py --speed-perturb true
     touch data/fbank/.primewords.done
   fi
 fi
@@ -180,7 +180,7 @@ if [ $stage -le 8 ] && [ $stop_stage -ge 8 ]; then
 
     if [ ! -f data/fbank/.magicdata.done ]; then
     mkdir -p data/fbank
-    ./local/compute_fbank_magicdata.py
+    ./local/compute_fbank_magicdata.py --speed-perturb true
     touch data/fbank/.magicdata.done
   fi
 fi
@@ -291,10 +291,10 @@ if [ $stage -le 12 ] && [ $stop_stage -ge 12 ]; then
     fi
 
     log "Compute KeSpeech fbank for train_phase1"
-    ./local/compute_fbank_kespeech_splits.py --num-splits ${num_splits} --training-subset train_phase1
+    ./local/compute_fbank_kespeech_splits.py --speed-perturb true --num-splits ${num_splits} --training-subset train_phase1
 
     log "Compute KeSpeech fbank for train_phase2"
-    ./local/compute_fbank_kespeech_splits.py --num-splits ${num_splits} --training-subset train_phase2
+    ./local/compute_fbank_kespeech_splits.py --speed-perturb true --num-splits ${num_splits} --training-subset train_phase2
 
     log "Compute KeSpeech fbank for test/dev"
     ./local/compute_fbank_kespeech_dev_test.py
@@ -344,10 +344,10 @@ if [ $stage -le 120 ] && [ $stop_stage -ge 120 ]; then
     fi
 
     log "Compute KeSpeech fbank for train_phase1"
-    ./local/compute_fbank_kespeech_splits.py --num-splits ${num_splits} --training-subset train_phase1 --num-mel-bins ${whisper_mel_bins} --whisper-fbank true
+    ./local/compute_fbank_kespeech_splits.py --speed-perturb true --num-splits ${num_splits} --training-subset train_phase1 --num-mel-bins ${whisper_mel_bins} --whisper-fbank true
 
     log "Compute KeSpeech fbank for train_phase2"
-    ./local/compute_fbank_kespeech_splits.py --num-splits ${num_splits} --training-subset train_phase2 --num-mel-bins ${whisper_mel_bins} --whisper-fbank true
+    ./local/compute_fbank_kespeech_splits.py --speed-perturb true --num-splits ${num_splits} --training-subset train_phase2 --num-mel-bins ${whisper_mel_bins} --whisper-fbank true
 
     log "Compute KeSpeech fbank for test/dev"
     ./local/compute_fbank_kespeech_dev_test.py --num-mel-bins ${whisper_mel_bins} --whisper-fbank true
@@ -356,19 +356,63 @@ if [ $stage -le 120 ] && [ $stop_stage -ge 120 ]; then
   fi
 fi
 
-if [ $stage -le 121 ] && [ $stop_stage -ge 121 ]; then
-  log "Stage 121: tmp"  
-    log "Compute KeSpeech fbank for train_phase1"
-    ./local/compute_fbank_kespeech_splits.py --num-splits ${num_splits} --stop 1 --training-subset train_phase1 --num-mel-bins ${whisper_mel_bins} --whisper-fbank true
+if [ $stage -le 122 ] && [ $stop_stage -ge 122 ]; then
+    log "Stage 122: Prepare speed perturb versionKeSpeech for whisper"
+    ./local/compute_fbank_kespeech_splits.py --speed-perturb true --num-splits ${num_splits} --training-subset train_phase1 --num-mel-bins ${whisper_mel_bins} --whisper-fbank true
 
     log "Compute KeSpeech fbank for train_phase2"
-    ./local/compute_fbank_kespeech_splits.py --num-splits ${num_splits} --training-subset train_phase2 --num-mel-bins ${whisper_mel_bins} --whisper-fbank true
+    ./local/compute_fbank_kespeech_splits.py --speed-perturb true --num-splits ${num_splits} --training-subset train_phase2 --num-mel-bins ${whisper_mel_bins} --whisper-fbank true
+fi
 
-    log "Compute KeSpeech fbank for test/dev"
-    ./local/compute_fbank_kespeech_dev_test.py --num-mel-bins ${whisper_mel_bins} --whisper-fbank true
+if [ $stage -le 121 ] && [ $stop_stage -ge 121 ]; then
+  log "Stage 121: Prepare MagicData, Primewords, ST-CMDS, THCHS-30 for whisper"
 
-    touch data/fbank/.kespeech.done
+  if [ ! -f data/manifests/.magicdata.done ]; then
+    mkdir -p data/manifests
+    lhotse prepare magicdata $dl_dir/magicdata data/manifests/magicdata
+    touch data/manifests/.magicdata.done
+  fi
+
+  if [ ! -f data/manifests/.primewords.done ]; then
+    mkdir -p data/manifests
+    lhotse prepare primewords $dl_dir/primewords data/manifests/primewords
+    touch data/manifests/.primewords.done
+  fi
+  if [ ! -f data/manifests/.stcmds.done ]; then
+    mkdir -p data/manifests
+    lhotse prepare stcmds $dl_dir/stcmds data/manifests/stcmds
+    touch data/manifests/.stcmds.done
+  fi
+
+  if [ ! -f data/manifests/.thchs30.done ]; then
+    mkdir -p data/manifests
+    lhotse prepare thchs-30 $dl_dir/thchs30 data/manifests/thchs30
+    touch data/manifests/.thchs30.done
+  fi
+
+  if [ ! -f data/fbank/.thchs30.done ]; then
+    mkdir -p data/fbank
+    ./local/compute_fbank_thchs30.py --speed-perturb true --num-mel-bins ${whisper_mel_bins} --whisper-fbank true
+    touch data/fbank/.thchs30.done
   fi
+
+  if [ ! -f data/fbank/.stcmds.done ]; then
+    mkdir -p data/fbank
+    ./local/compute_fbank_stcmds.py --speed-perturb true --num-mel-bins ${whisper_mel_bins} --whisper-fbank true
+    touch data/fbank/.stcmds.done
+  fi
+  if [ ! -f data/fbank/.magicdata.done ]; then
+    mkdir -p data/fbank
+    ./local/compute_fbank_magicdata.py --speed-perturb true --num-mel-bins ${whisper_mel_bins} --whisper-fbank true
+    touch data/fbank/.magicdata.done
+  fi
+
+  if [ ! -f data/fbank/.primewords.done ]; then
+    mkdir -p data/fbank
+    ./local/compute_fbank_primewords.py --speed-perturb true --num-mel-bins ${whisper_mel_bins} --whisper-fbank true
+    touch data/fbank/.primewords.done
+  fi
+
 fi