Check that read_manifests_if_cached returns a non-empty dict.

egs/aidatatang_200zh/ASR/local/compute_fbank_aidatatang_200zh.py

@@ -62,6 +62,13 @@ def compute_fbank_aidatatang_200zh(num_mel_bins: int = 80):
     )
     assert manifests is not None
 
+    assert len(manifests) == len(dataset_parts), (
+        len(manifests),
+        len(dataset_parts),
+        list(manifests.keys()),
+        dataset_parts,
+    )
+
     extractor = Fbank(FbankConfig(num_mel_bins=num_mel_bins))
 
     with get_executor() as ex:  # Initialize the executor only once.

egs/aishell/ASR/local/compute_fbank_aidatatang_200zh.py

@@ -62,6 +62,13 @@ def compute_fbank_aidatatang_200zh(num_mel_bins: int = 80):
     )
     assert manifests is not None
 
+    assert len(manifests) == len(dataset_parts), (
+        len(manifests),
+        len(dataset_parts),
+        list(manifests.keys()),
+        dataset_parts,
+    )
+
     extractor = Fbank(FbankConfig(num_mel_bins=num_mel_bins))
 
     with get_executor() as ex:  # Initialize the executor only once.

egs/aishell/ASR/local/compute_fbank_aishell.py

@@ -62,6 +62,13 @@ def compute_fbank_aishell(num_mel_bins: int = 80):
     )
     assert manifests is not None
 
+    assert len(manifests) == len(dataset_parts), (
+        len(manifests),
+        len(dataset_parts),
+        list(manifests.keys()),
+        dataset_parts,
+    )
+
     extractor = Fbank(FbankConfig(num_mel_bins=num_mel_bins))
 
     with get_executor() as ex:  # Initialize the executor only once.

egs/aishell2/ASR/local/compute_fbank_aishell2.py

@@ -62,6 +62,13 @@ def compute_fbank_aishell2(num_mel_bins: int = 80):
     )
     assert manifests is not None
 
+    assert len(manifests) == len(dataset_parts), (
+        len(manifests),
+        len(dataset_parts),
+        list(manifests.keys()),
+        dataset_parts,
+    )
+
     extractor = Fbank(FbankConfig(num_mel_bins=num_mel_bins))
 
     with get_executor() as ex:  # Initialize the executor only once.

egs/aishell4/ASR/local/compute_fbank_aishell4.py

@@ -63,6 +63,13 @@ def compute_fbank_aishell4(num_mel_bins: int = 80):
     )
     assert manifests is not None
 
+    assert len(manifests) == len(dataset_parts), (
+        len(manifests),
+        len(dataset_parts),
+        list(manifests.keys()),
+        dataset_parts,
+    )
+
     extractor = Fbank(FbankConfig(num_mel_bins=num_mel_bins))
 
     with get_executor() as ex:  # Initialize the executor only once.

egs/alimeeting/ASR/local/compute_fbank_alimeeting.py

@@ -63,6 +63,13 @@ def compute_fbank_alimeeting(num_mel_bins: int = 80):
     )
     assert manifests is not None
 
+    assert len(manifests) == len(dataset_parts), (
+        len(manifests),
+        len(dataset_parts),
+        list(manifests.keys()),
+        dataset_parts,
+    )
+
     extractor = Fbank(FbankConfig(num_mel_bins=num_mel_bins))
 
     with get_executor() as ex:  # Initialize the executor only once.

egs/gigaspeech/ASR/local/preprocess_gigaspeech.py

@@ -62,6 +62,13 @@ def preprocess_giga_speech():
     )
     assert manifests is not None
 
+    assert len(manifests) == len(dataset_parts), (
+        len(manifests),
+        len(dataset_parts),
+        list(manifests.keys()),
+        dataset_parts,
+    )
+
     for partition, m in manifests.items():
         logging.info(f"Processing {partition}")
         raw_cuts_path = output_dir / f"cuts_{partition}_raw.jsonl.gz"

egs/librispeech/ASR/local/compute_fbank_librispeech.py

@@ -66,6 +66,13 @@ def compute_fbank_librispeech():
     )
     assert manifests is not None
 
+    assert len(manifests) == len(dataset_parts), (
+        len(manifests),
+        len(dataset_parts),
+        list(manifests.keys()),
+        dataset_parts,
+    )
+
     extractor = Fbank(FbankConfig(num_mel_bins=num_mel_bins))
 
     with get_executor() as ex:  # Initialize the executor only once.

egs/librispeech/ASR/local/compute_fbank_musan.py

@@ -65,6 +65,8 @@ def compute_fbank_musan():
     assert len(manifests) == len(dataset_parts), (
         len(manifests),
         len(dataset_parts),
+        list(manifests.keys()),
+        dataset_parts,
     )
 
     musan_cuts_path = output_dir / "musan_cuts.jsonl.gz"

egs/librispeech/ASR/local/preprocess_gigaspeech.py

@@ -68,6 +68,13 @@ def preprocess_giga_speech():
     )
     assert manifests is not None
 
+    assert len(manifests) == len(dataset_parts), (
+        len(manifests),
+        len(dataset_parts),
+        list(manifests.keys()),
+        dataset_parts,
+    )
+
     for partition, m in manifests.items():
         logging.info(f"Processing {partition}")
         raw_cuts_path = output_dir / f"{prefix}_cuts_{partition}_raw.{suffix}"

egs/spgispeech/ASR/local/compute_fbank_musan.py

@@ -69,6 +69,13 @@ def compute_fbank_musan():
     )
     assert manifests is not None
 
+    assert len(manifests) == len(dataset_parts), (
+        len(manifests),
+        len(dataset_parts),
+        list(manifests.keys()),
+        dataset_parts,
+    )
+
     musan_cuts_path = src_dir / "cuts_musan.jsonl.gz"
 
     if musan_cuts_path.is_file():

egs/tal_csasr/ASR/local/compute_fbank_tal_csasr.py

@@ -62,6 +62,13 @@ def compute_fbank_tal_csasr(num_mel_bins: int = 80):
     )
     assert manifests is not None
 
+    assert len(manifests) == len(dataset_parts), (
+        len(manifests),
+        len(dataset_parts),
+        list(manifests.keys()),
+        dataset_parts,
+    )
+
     extractor = Fbank(FbankConfig(num_mel_bins=num_mel_bins))
 
     with get_executor() as ex:  # Initialize the executor only once.

egs/tedlium3/ASR/local/compute_fbank_tedlium.py

@@ -62,6 +62,13 @@ def compute_fbank_tedlium():
     )
     assert manifests is not None
 
+    assert len(manifests) == len(dataset_parts), (
+        len(manifests),
+        len(dataset_parts),
+        list(manifests.keys()),
+        dataset_parts,
+    )
+
     extractor = Fbank(FbankConfig(num_mel_bins=num_mel_bins))
 
     with get_executor() as ex:  # Initialize the executor only once.

egs/timit/ASR/local/compute_fbank_timit.py

@@ -63,6 +63,13 @@ def compute_fbank_timit():
     )
     assert manifests is not None
 
+    assert len(manifests) == len(dataset_parts), (
+        len(manifests),
+        len(dataset_parts),
+        list(manifests.keys()),
+        dataset_parts,
+    )
+
     extractor = Fbank(FbankConfig(num_mel_bins=num_mel_bins))
 
     with get_executor() as ex:  # Initialize the executor only once.

egs/wenetspeech/ASR/local/preprocess_wenetspeech.py

@@ -72,6 +72,13 @@ def preprocess_wenet_speech():
     )
     assert manifests is not None
 
+    assert len(manifests) == len(dataset_parts), (
+        len(manifests),
+        len(dataset_parts),
+        list(manifests.keys()),
+        dataset_parts,
+    )
+
     for partition, m in manifests.items():
         logging.info(f"Processing {partition}")
         raw_cuts_path = output_dir / f"cuts_{partition}_raw.jsonl.gz"

egs/yesno/ASR/local/compute_fbank_yesno.py

@@ -47,6 +47,13 @@ def compute_fbank_yesno():
     )
     assert manifests is not None
 
+    assert len(manifests) == len(dataset_parts), (
+        len(manifests),
+        len(dataset_parts),
+        list(manifests.keys()),
+        dataset_parts,
+    )
+
     extractor = Fbank(
         FbankConfig(sampling_rate=8000, num_mel_bins=num_mel_bins)
     )