[update] update readme and add cli usage for the dataset

README.md

@@ -1,2 +1,40 @@
 # Whisper-lyric
+
 Codebase to finetune whisper for music transcription.
+
+## Installation
+
+To install the required dependencies, run the following command:
+
+```bash
+pip install -r requirements.txt
+```
+
+## Dataset download
+
+To download the dataset, run the following command:
+
+```bash
+python download_dataset.py --num_images 1000
+```
+
+The dataset will be downloaded to the `data` directory.
+The format of the dataset is as follows:
+
+```
+dataset
+├── audio
+│ ├── 0.wav
+│ ├── 1.wav
+│ ├── ...
+└── lyrics
+ ├── 0.txt
+ ├── 1.txt
+ ├── ...
+```
+
+where `0.wav` corresponds to the audio file and `0.txt` corresponds to the lyrics transcription of the audio file.
+
+## License
+
+This project is licensed under the MIT License - see the [LICENSE](LICENSE) file for details.

dataset/build.py

@@ -6,7 +6,6 @@
 import orjson
 import requests as r
 
-
 _BASE_URL = (
  "https://firestore.googleapis.com/google.firestore.v1.Firestore/Listen/channel"
 )
@@ -169,6 +168,8 @@ def _construct_ds(self, audio_data: dict) -> None:
  with open(f"./dataset/audio/{nbm_file}.ogg", "wb") as file:
  file.write(resp.content)
 
+ print(f"Downloaded {nbm_file} songs", end="\r")
+
  def download(self, nb_element: int = 10) -> None:
  """Method to download the data
 

dataset_builder.py

@@ -0,0 +1,15 @@
+import argparse
+
+from dataset.build import SonautoAPI
+
+
+parser = argparse.ArgumentParser(
+ description="Download images from Sonauto dataset",
+)
+parser.add_argument("--num_images", type=int, default=10)
+parser.add_argument("--clean", type=bool, default=True)
+
+args = parser.parse_args()
+
+api = SonautoAPI(clean=args.clean)
+api.download(args.num_images)