nit refactoring

download_audio.py

@@ -4,13 +4,18 @@ import tarfile
 import zipfile
 import gzip
 import subprocess
+import time
 from os.path import join as p_join
 from tqdm import tqdm
 from multiprocessing import Pool
-from typing import Optional
+from typing import Optional, Dict
+from glob import glob
 
 import pandas as pd
+import soundfile as sf
+from datasets import Dataset, Audio, DatasetDict
 
+audio_loader = Audio()
 # dataset config
 url_metadata_dict = {
     "enA-jaA": "https://dl.fbaipublicfiles.com/seamless/data/seamless_align_nov2023_extension/seamless.dataset.metadata.public.enA-jaA.tsv.gz",
@@ -25,10 +30,14 @@ for s in sides:
     os.makedirs(p_join(cache_dir_audio, s), exist_ok=True)
 # processor config
 n_pool = int(os.getenv("N_POOL", 8))
-wget_max_retry = os.getenv("MAX_RETRY", "1")
-wget_timeout = os.getenv("TIMEOUT", "20")
+wget_max_retry = os.getenv("MAX_RETRY", "2")
+wget_timeout = os.getenv("TIMEOUT", "30")
 line_no_start = int(os.getenv("LINE_NO_START", 0))
 line_no_end = int(os.getenv("LINE_NO_END", 10000))
+dataset_id = int(os.getenv("DATASET_ID", 0))
+hf_org = "kotoba-tech"
+hf_dataset = f"seamless-align-{direction}"
+
 
 
 def wget(url: str, output_file: Optional[str] = None):
@@ -87,10 +96,23 @@ def get_audio(dataframe: pd.DataFrame):
         features.update({f"{side}.{k}": to_json_serializable(v) for k, v in df.iloc[0].to_dict().items()})
         identifier = os.path.basename(features[f"{side}.url"]).split(".")[-1]
         features[f"{side}.path"] = str(p_join(cache_dir_audio, side, f"{features['line_no']}.{identifier}"))
+        start, end = features[f"{side}.duration_start"], features[f"{side}.duration_end"]
         if not os.path.exists(features[f"{side}.path"]):
             flag = wget(features[f"{side}.url"], output_file=features[f"{side}.path"])
             if not flag:
                 return False
+            else:
+                try:
+                    wav = audio_loader.decode_example({"path": features[f"{side}.path"], "bytes": None})
+                    if start < end < len(wav["array"]):
+                        sf.write(features[f"{side}.path"], wav["array"][start:end], wav["sampling_rate"])
+                    else:
+                        os.remove(features[f"{side}.path"])
+                        return False
+                except Exception as e:
+                    print(e)
+                    os.remove(features[f"{side}.path"])
+                    return False
     with open(p_join(cache_dir_feature, f'{features["line_no"]}.json'), "w") as f:
         json.dump(features, f)
     return True
@@ -119,6 +141,30 @@ def process_dataset():
             pool.map(get_audio, tqdm(inputs, total=len(inputs)))
 
 
+    def loader(feature: str) -> Dict:
+        with open(feature) as f_reader:
+            return json.load(f_reader)
+
+    features = [loader(i) for i in glob(p_join(cache_dir_feature, '*.json'))]
+    print(f"push {len(features)} records to hub")
+    data_dict = {}
+    for side in sides:
+        data_dict.update({f"{side}.audio": [i.pop(f"{side}.path") for i in features]})
+    data_dict.update({k: [i[k] for i in features] for k in features[0].keys()})
+    audio_dataset = Dataset.from_dict(data_dict)
+    for side in sides:
+        audio_dataset = audio_dataset.cast_column(f"{side}.audio", Audio())
+    dataset_to_push = DatasetDict({"train": audio_dataset})
+    repo_name = f"{hf_org}/{hf_dataset}"
+    while True:
+        try:
+            dataset_to_push.push_to_hub(repo_name, config_name=f"subset_{dataset_id}")
+            break
+        except Exception:
+            print(f"FAILED: push_to_hub on {repo_name} failed. wait 60 sec and retry soon...")
+            time.sleep(60)
+
+
 if __name__ == '__main__':
     process_dataset()
 

filter_audio.py

@@ -48,4 +48,4 @@ for i in tqdm(index_list):
                 sf.write(p_join(cache_dir_audio_fixed, f"{i}.mp3"), wav["array"][start:end], wav["sampling_rate"])
         except Exception as e:
             print(e)
-        os.remove(audio_file)
+            os.remove(audio_file)

main.sh

@@ -2,82 +2,134 @@
 # enA-jaA: 718_606 #
 ####################
 export DIRECTION="enA-jaA"
-
-export DATASET_ID=1
 export LINE_NO_START=0
 export LINE_NO_END=25000
 python download_audio.py
 
-export DATASET_ID=2
+export DIRECTION="enA-jaA"
 export LINE_NO_START=25000
 export LINE_NO_END=50000
 python download_audio.py
 
-export DATASET_ID=3
+export DIRECTION="enA-jaA"
 export LINE_NO_START=50000
 export LINE_NO_END=75000
 python download_audio.py
 
-export DATASET_ID=4
+export DIRECTION="enA-jaA"
 export LINE_NO_START=75000
 export LINE_NO_END=100000
 python download_audio.py
 
-export DATASET_ID=5
+export DIRECTION="enA-jaA"
 export LINE_NO_START=100000
 export LINE_NO_END=125000
 python download_audio.py
 
-export DATASET_ID=6
+export DIRECTION="enA-jaA"
 export LINE_NO_START=125000
 export LINE_NO_END=150000
 python download_audio.py
 
-export DATASET_ID=7
+export DIRECTION="enA-jaA"
 export LINE_NO_START=150000
 export LINE_NO_END=175000
 python download_audio.py
 
-export DATASET_ID=8
+export DIRECTION="enA-jaA"
 export LINE_NO_START=175000
 export LINE_NO_END=200000
 python download_audio.py
 
-export DATASET_ID=9
+export DIRECTION="enA-jaA"
 export LINE_NO_START=200000
 export LINE_NO_END=225000
 python download_audio.py
 
-export DATASET_ID=10
+export DIRECTION="enA-jaA"
 export LINE_NO_START=225000
 export LINE_NO_END=250000
 python download_audio.py
 
-export DATASET_ID=11
+export DIRECTION="enA-jaA"
 export LINE_NO_START=250000
 export LINE_NO_END=275000
 python download_audio.py
 
-export DATASET_ID=12
+export DIRECTION="enA-jaA"
 export LINE_NO_START=275000
 export LINE_NO_END=300000
 python download_audio.py
 
-export DATASET_ID=13
+export DIRECTION="enA-jaA"
 export LINE_NO_START=300000
 export LINE_NO_END=325000
 python download_audio.py
 
-export DATASET_ID=14
+export DIRECTION="enA-jaA"
 export LINE_NO_START=325000
 export LINE_NO_END=350000
 python download_audio.py
 
-export DATASET_ID=15
+export DIRECTION="enA-jaA"
 export LINE_NO_START=350000
 export LINE_NO_END=360000
 python download_audio.py
-python push_s2s_translation.py
+
+# FILTER AUDIO
+export DIRECTION="enA-jaA"
+export DIRECTION_SPEECH="enA"
+export LINE_NO_START=0
+export LINE_NO_END=25000
+python filter_audio.py
+
+export DIRECTION="enA-jaA"
+export DIRECTION_SPEECH="enA"
+export LINE_NO_START=25000
+export LINE_NO_END=50000
+python filter_audio.py
+
+export DIRECTION="enA-jaA"
+export DIRECTION_SPEECH="enA"
+export LINE_NO_START=50000
+export LINE_NO_END=75000
+python filter_audio.py
+
+export DIRECTION="enA-jaA"
+export DIRECTION_SPEECH="enA"
+export LINE_NO_START=75000
+export LINE_NO_END=100000
+python filter_audio.py
+
+export DIRECTION="enA-jaA"
+export DIRECTION_SPEECH="enA"
+export LINE_NO_START=100000
+export LINE_NO_END=125000
+python filter_audio.py
+
+export DIRECTION="enA-jaA"
+export DIRECTION_SPEECH="enA"
+export LINE_NO_START=125000
+export LINE_NO_END=150000
+python filter_audio.py
+
+export DIRECTION="enA-jaA"
+export DIRECTION_SPEECH="enA"
+export LINE_NO_START=150000
+export LINE_NO_END=175000
+python filter_audio.py
+
+export DIRECTION="enA-jaA"
+export DIRECTION_SPEECH="enA"
+export LINE_NO_START=175000
+export LINE_NO_END=200000
+python filter_audio.py
+
+export DIRECTION="enA-jaA"
+export DIRECTION_SPEECH="enA"
+export LINE_NO_START=200000
+export LINE_NO_END=225000
+python filter_audio.py
 
 ######################
 # enA-jpn: 1_468_292 #