Upload README.md with huggingface_hub

README.md

@@ -1,31 +1,259 @@
----
-dataset_info:
-  features:
-  - name: audio_filepath
-    dtype: string
-  - name: text
-    dtype: string
-  - name: duration
-    dtype: float64
-  splits:
-  - name: train
-    num_bytes: 346388747
-    num_examples: 1562732
-  - name: valid
-    num_bytes: 19221002
-    num_examples: 86814
-  - name: test
-    num_bytes: 19216212
-    num_examples: 86827
-  download_size: 134440031
-  dataset_size: 384825961
-configs:
-- config_name: default
-  data_files:
-  - split: train
-    path: data/train-*
-  - split: valid
-    path: data/valid-*
-  - split: test
-    path: data/test-*
----
+# Meta Speech Recognition Slavic Languages Dataset (Common Voice)
+
+This dataset contains metadata for Slavic language speech recognition samples from Common Voice.
+
+## Dataset Sources and Credits
+
+This dataset contains samples from Mozilla Common Voice:
+- Source: https://commonvoice.mozilla.org/en/datasets
+- License: CC0-1.0
+- Citation: Please acknowledge Mozilla Common Voice if you use this data
+
+## Languages Included
+
+The dataset includes the following Slavic languages:
+- Belarusian (be)
+- Bulgarian (bg)
+- Czech (cs)
+- Georgian (ka)
+- Macedonian (mk)
+- Polish (pl)
+- Russian (ru)
+- Slovak (sk)
+- Slovenian (sl)
+- Serbian (sr)
+- Ukrainian (uk)
+
+## Dataset Statistics
+
+### Splits and Sample Counts
+- **train**: 1562732 samples
+- **valid**: 86814 samples
+- **test**: 86827 samples
+
+## Example Samples
+### train
+```json
+{
+  "audio_filepath": "/cv/cv-corpus-15.0-2023-09-08/be/clips/common_voice_be_35310612.mp3",
+  "text": "Аднак электарат вырашыў інакш. AGE_18_30 GER_MALE EMOTION_NEUTRAL INTENT_INFORM",
+  "duration": 5.22
+}
+```
+```json
+{
+  "audio_filepath": "/cv/cv-corpus-15.0-2023-09-08/be/clips/common_voice_be_32521083.mp3",
+  "text": "Але гэта не адзіны паток. AGE_18_30 GER_FEMALE EMOTION_NEUTRAL INTENT_INFORM",
+  "duration": 2.74
+}
+```
+
+### valid
+```json
+{
+  "audio_filepath": "/cv/cv-corpus-15.0-2023-09-08/be/clips/common_voice_be_29218733.mp3",
+  "text": "Вы ж зразумейце наколькі гэта складана. AGE_18_30 GER_MALE EMOTION_NEUTRAL INTENT_INFORM",
+  "duration": 4.32
+}
+```
+```json
+{
+  "audio_filepath": "/cv/cv-corpus-15.0-2023-09-08/be/clips/common_voice_be_29003430.mp3",
+  "text": "Адмовіцца ад яго немагчыма. AGE_18_30 GER_MALE EMOTION_NEUTRAL INTENT_ASSERT",
+  "duration": 3.42
+}
+```
+
+### test
+```json
+{
+  "audio_filepath": "/cv/cv-corpus-15.0-2023-09-08/be/clips/common_voice_be_29742537.mp3",
+  "text": "Гарэла святло ў хатах. AGE_18_30 GER_FEMALE EMOTION_FEAR INTENT_INFORM",
+  "duration": 2.45
+}
+```
+```json
+{
+  "audio_filepath": "/cv/cv-corpus-15.0-2023-09-08/be/clips/common_voice_be_28447241.mp3",
+  "text": "А новых грошай узяць няма адкуль. AGE_18_30 GER_MALE EMOTION_NEUTRAL INTENT_INFORM",
+  "duration": 3.31
+}
+```
+
+
+## Downloading Audio Files
+
+To use this dataset, you need to download the Common Voice dataset for each language. The audio files are not included in this repository.
+
+### Download Instructions
+
+1. Download Common Voice dataset version 15.0 (2023-09-08) for each language:
+```bash
+# For each language (replace {lang} with language code: be, bg, cs, ka, mk, pl, ru, sk, sl, sr, uk)
+wget https://voice-prod-bundler-ee1969a6ce8178826482b88e843c335139bd3fb4.s3.amazonaws.com/cv-corpus-15.0-2023-09-08/{lang}.tar.gz
+
+# Extract the files
+tar -xzf {lang}.tar.gz
+```
+
+2. Place the extracted files in the following structure:
+```
+/cv/cv-corpus-15.0-2023-09-08/
+├── be/
+├── bg/
+├── cs/
+├── ka/
+├── mk/
+├── pl/
+├── ru/
+├── sk/
+├── sl/
+├── sr/
+└── uk/
+```
+
+## Training NeMo Conformer ASR for Slavic Languages
+
+### 1. Pull and Run NeMo Docker
+```bash
+# Pull the NeMo Docker image
+docker pull nvcr.io/nvidia/nemo:24.05
+
+# Run the container with GPU support
+docker run --gpus all -it --rm \
+    -v /external1:/external1 \
+    -v /external2:/external2 \
+    -v /external3:/external3 \
+    -v /cv:/cv \
+    --shm-size=8g \
+    -p 8888:8888 -p 6006:6006 \
+    --ulimit memlock=-1 \
+    --ulimit stack=67108864 \
+    nvcr.io/nvidia/nemo:24.05
+```
+
+### 2. Create Training Script
+Create a script `train_nemo_asr_slavic.py`:
+```python
+from nemo.collections.asr.models import EncDecCTCModel
+from nemo.collections.asr.data.audio_to_text import TarredAudioToTextDataset
+import pytorch_lightning as pl
+from omegaconf import OmegaConf
+import os
+
+# Load the dataset from Hugging Face
+from datasets import load_dataset
+dataset = load_dataset("WhissleAI/Meta_STT_SLAVIC_CommonVoice")
+
+# Create config
+config = OmegaConf.create({
+    'model': {
+        'name': 'EncDecCTCModel',
+        'train_ds': {
+            'manifest_filepath': None,  # Will be set dynamically
+            'batch_size': 32,
+            'shuffle': True,
+            'num_workers': 4,
+            'pin_memory': True,
+            'use_start_end_token': False,
+        },
+        'validation_ds': {
+            'manifest_filepath': None,  # Will be set dynamically
+            'batch_size': 32,
+            'shuffle': False,
+            'num_workers': 4,
+            'pin_memory': True,
+            'use_start_end_token': False,
+        },
+        'optim': {
+            'name': 'adamw',
+            'lr': 0.001,
+            'weight_decay': 0.01,
+        },
+        'trainer': {
+            'devices': 1,
+            'accelerator': 'gpu',
+            'max_epochs': 100,
+            'precision': 16,
+        }
+    }
+})
+
+# Initialize model
+model = EncDecCTCModel(cfg=config.model)
+
+# Create trainer
+trainer = pl.Trainer(**config.model.trainer)
+
+# Train
+trainer.fit(model)
+```
+
+### 3. Create Config File
+Create a config file `config_slavic.yaml`:
+```yaml
+model:
+  name: "EncDecCTCModel"
+  train_ds:
+    manifest_filepath: "train.json"
+    batch_size: 32
+    shuffle: true
+    num_workers: 4
+    pin_memory: true
+    use_start_end_token: false
+    
+  validation_ds:
+    manifest_filepath: "valid.json"
+    batch_size: 32
+    shuffle: false
+    num_workers: 4
+    pin_memory: true
+    use_start_end_token: false
+    
+  optim:
+    name: adamw
+    lr: 0.001
+    weight_decay: 0.01
+    
+  trainer:
+    devices: 1
+    accelerator: "gpu"
+    max_epochs: 100
+    precision: 16
+```
+
+### 4. Start Training
+```bash
+# Inside the NeMo container
+python -m torch.distributed.launch --nproc_per_node=1 \
+    train_nemo_asr_slavic.py \
+    --config-path=. \
+    --config-name=config_slavic.yaml
+```
+
+## Usage Notes
+
+1. The dataset includes only metadata. Audio files must be downloaded separately from Common Voice.
+2. Audio files should be placed in the `/cv/cv-corpus-15.0-2023-09-08/{lang}/` directory structure.
+3. For optimal performance:
+   - Use a GPU with at least 16GB VRAM
+   - Adjust batch size based on your GPU memory
+   - Consider gradient accumulation for larger effective batch sizes
+   - Monitor training with TensorBoard (accessible via port 6006)
+
+## Common Issues and Solutions
+
+1. **Memory Issues**: 
+   - Reduce batch size if you encounter OOM errors
+   - Use gradient accumulation for larger effective batch sizes
+   - Enable mixed precision training (fp16)
+
+2. **Training Speed**:
+   - Increase num_workers based on your CPU cores
+   - Use pin_memory=True for faster data transfer to GPU
+   - Consider using tarred datasets for faster I/O
+
+3. **Model Performance**:
+   - Adjust learning rate based on your batch size
+   - Use learning rate warmup for better convergence
+   - Consider using a pretrained model as initialization