init

main_s2s.sh

@@ -84,7 +84,7 @@ do
   python fetch_dataset_s2s.py
 done
 # tokenize
-for i in $(seq 81 100);
+for i in $(seq 101 150);
 do
   export N_POOL=15
   export DATASET_ID=${i}
@@ -95,7 +95,7 @@ do
   python tokenize_dataset_s2s.py
 done
 # speaker embedding
-for i in $(seq 81 100);
+for i in $(seq 101 150);
 do
   export DATASET_ID=${i}
   export DIRECTION="enA-zhA"
@@ -132,7 +132,7 @@ do
   python fetch_dataset_s2s.py
 done
 # tokenize
-for i in $(seq 1 30);
+for i in $(seq 31 60);
 do
   export N_POOL=15
   export DATASET_ID=${i}
@@ -142,7 +142,7 @@ do
   echo ${LINE_NO_START}
   python tokenize_dataset_s2s.py
 done
-for i in $(seq 300 330);
+for i in $(seq 331 360);
 do
   export N_POOL=15
   export DATASET_ID=${i}
@@ -153,14 +153,14 @@ do
   python tokenize_dataset_s2s.py
 done
 # speaker embedding
-for i in $(seq 1 30);
+for i in $(seq 31 60);
 do
   export DATASET_ID=${i}
   export DIRECTION="enA-frA"
   python attach_speaker_embedding_s2s.py
 done
 # speaker embedding
-for i in $(seq 300 330);
+for i in $(seq 331 360);
 do
   export DATASET_ID=${i}
   export DIRECTION="enA-frA"
@@ -197,7 +197,19 @@ do
   python fetch_dataset_s2s.py
 done
 # tokenize
-for i in $(seq 1 20);
+#for i in 215 216 232;
+#do
+#  export N_POOL=15
+#  export DATASET_ID=${i}
+#  export DIRECTION="deA-enA"
+#  export LINE_NO_START=$(((DATASET_ID-1) * 2500))
+#  export LINE_NO_END=$((DATASET_ID * 2500))
+#  echo ${LINE_NO_START}
+#  python tokenize_dataset_s2s.py
+#
+#done
+
+for i in $(seq 21 50);
 do
   export N_POOL=15
   export DATASET_ID=${i}
@@ -207,7 +219,7 @@ do
   echo ${LINE_NO_START}
   python tokenize_dataset_s2s.py
 done
-for i in $(seq 200 240);
+for i in $(seq 241 290);
 do
   export N_POOL=15
   export DATASET_ID=${i}
@@ -218,14 +230,14 @@ do
   python tokenize_dataset_s2s.py
 done
 # speaker embedding
-for i in $(seq 1 20);
+for i in $(seq 21 50);
 do
   export DATASET_ID=${i}
   export DIRECTION="deA-enA"
   python attach_speaker_embedding_s2s.py
 done
 # speaker embedding
-for i in $(seq 200 240);
+for i in $(seq 241 290);
 do
   export DATASET_ID=${i}
   export DIRECTION="deA-enA"

tokenize_dataset_s2s.py

@@ -23,7 +23,9 @@ audio_loader = Audio()
 def error_file(example):
     for side in sides:
         try:
-            audio_loader.decode_example(example[f"{side}.audio"])
+            wav = audio_loader.decode_example(example[f"{side}.audio"])
+            if len(wav["array"]) == 0:
+                return False
         except LibsndfileError:
             return False
     return True
@@ -41,6 +43,8 @@ print(f"Num examples (after filtering): {len(dataset)}")
 def tokenize(example):
     for side in sides:
         wav = torch.as_tensor(example[f"{side}.audio"]["array"].reshape(1, 1, -1), dtype=torch.float32)
+        if len(wav) == 0:
+            return None
         example[f"{side}.audio.tokens"] = tokenizer.wav_to_tokens(
             wav=wav, sample_rate=example[f"{side}.audio"]["sampling_rate"]
         ).numpy().tolist()[0]