Upload README.md with huggingface_hub

README.md

@@ -1,32 +1,224 @@
----
-license: cc-by-4.0
-configs:
-- config_name: default
-  data_files:
-  - split: train
-    path: data/train-*
-  - split: valid
-    path: data/valid-*
-  - split: test
-    path: data/test-*
-dataset_info:
-  features:
-  - name: audio_filepath
-    dtype: audio
-  - name: text
-    dtype: string
-  - name: duration
-    dtype: float64
-  splits:
-  - name: train
-    num_bytes: 433693291547.502
-    num_examples: 759309
-  - name: valid
-    num_bytes: 11490209888.704
-    num_examples: 37464
-  - name: test
-    num_bytes: 23113506775.214
-    num_examples: 43406
-  download_size: 155500308826
-  dataset_size: 468297008211.42
----
+# Meta Speech Recognition Hindi Dataset (Set 1)
+
+This dataset contains both metadata and audio files for Hindi speech recognition samples, curated from multiple sources.
+
+## Dataset Sources and Credits
+
+This dataset combines samples from the following sources:
+
+1. **AI4Bharat Indic Speech Dataset**
+   - Source: https://ai4bharat.org/indic-speech-dataset
+   - License: CC-BY 4.0
+   - Citation: Please cite the original paper if you use this data
+
+2. **Common Voice Hindi**
+   - Source: https://commonvoice.mozilla.org/en/datasets
+   - License: CC0-1.0
+   - Citation: Please acknowledge Mozilla Common Voice if you use this data
+
+3. **Shrutilipi**
+   - Source: https://ai4bharat.org/shrutilipi
+   - License: CC-BY 4.0
+   - Citation: Please cite the original paper if you use this data
+
+## Dataset Statistics
+
+### Splits and Sample Counts
+- **train**: 759309 samples
+- **valid**: 37464 samples
+- **test**: 43406 samples
+
+## Example Samples
+### train
+```json
+{
+  "audio_filepath": "/external4/datasets/AI4Bharat/hindi/wavs_train/5348024557584410_0.wav",
+  "text": "पालन INTENT_KEYWORDS_SPOTTING AGE_60+ GENDER_MALE DIALECT_RAJASTHAN",
+  "duration": 1.32
+}
+```
+```json
+{
+  "audio_filepath": "/external4/datasets/AI4Bharat/hindi/wavs_train/5348024557795852_0.wav",
+  "text": "हमारी मातृभाषा ENTITY_LANGUAGE हिंदी END यह इतनी दिलचस्प है कि किसी को ENTITY_TIME दिन भर END में बोलने के लिए हमें कई प्रकार के मुहावरों का इस्तेमाल किया जाता है INTENT_LANGUAGE_SPECIFIC AGE_30_45 GENDER_MALE DIALECT_MADHYA_PRADESH",
+  "duration": 10.816
+}
+```
+
+### valid
+```json
+{
+  "audio_filepath": "/external4/datasets/AI4Bharat/hindi/wavs/8012ebba-be5d-4cf1-953d-1282fe7d75c4_0_1.wav",
+  "text": "ENTITY_ORGANIZATION जी ट्रैवलिंग ट्रैवल एजेंसी END से बोल रए हैं सर INTENT_CONVERSATION AGE_30_45 GENDER_MALE DIALECT_MADHYA_PRADESH",
+  "duration": 2.9440000000000004
+}
+```
+```json
+{
+  "audio_filepath": "/external4/datasets/AI4Bharat/hindi/wavs/8012ebba-be5d-4cf1-953d-1282fe7d75c4_0_2.wav",
+  "text": "मुझे ENTITY_PERSON_NAME सर ओमकारेश्‍वर END के लिए ENTITY_VEHICLE_TYPE सेवेन सीटर गाड़ी END चहिए थी INTENT_CONVERSATION AGE_30_45 GENDER_MALE DIALECT_MADHYA_PRADESH",
+  "duration": 3.808
+}
+```
+
+### test
+```json
+{
+  "audio_filepath": "/external4/datasets/AI4Bharat/hindi/wavs_train/82ff7f7c-8c24-4ff2-8c99-74149ab6016b_1_25.wav",
+  "text": "त दु सौ के लीजिए देखिए पहिला बोहनी है आपका करिए INTENT_CONVERSATION AGE_45_60 GENDER_FEMALE DIALECT_BIHAR",
+  "duration": 4.289999999999992
+}
+```
+```json
+{
+  "audio_filepath": "/external4/datasets/AI4Bharat/hindi/wavs_train/82ff7f7c-8c24-4ff2-8c99-74149ab6016b_1_26.wav",
+  "text": "हां त पहला बोहनी है आप खरीद लीजिए हमको ओ टोक दिए हैं और जाइए नहीं बिकेगा तो उ भी हमको बहुत बती होता है जाना आप ले लीजिए न INTENT_CONVERSATION AGE_45_60 GENDER_FEMALE DIALECT_BIHAR",
+  "duration": 11.733000000000004
+}
+```
+
+
+## Training NeMo Conformer ASR for Hindi
+
+### 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 \
+    --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_hindi.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_HI_Set1")
+
+# 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_hindi.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_hindi.py \
+    --config-path=. \
+    --config-name=config_hindi.yaml
+```
+
+## Usage Notes
+
+1. The dataset includes both metadata and audio files.
+2. Audio files are stored in the dataset repository.
+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