Upload dataset as Parquet (1 files, 491 records)

README.md

@@ -0,0 +1,160 @@
+---
+license: cc-by-4.0
+task_categories:
+- automatic-speech-recognition
+- text-to-speech
+language:
+- tr
+tags:
+- speech
+- audio
+- dataset
+- tts
+- asr
+- merged-dataset
+size_categories:
+- n<1K
+configs:
+- config_name: default
+  data_files:
+  - split: train
+    path: "data.jsonl"
+  default: true
+dataset_info:
+  features:
+  - name: audio
+    dtype:
+      audio:
+        sampling_rate: null
+  - name: text
+    dtype: string
+  - name: speaker_id
+    dtype: string
+  - name: emotion
+    dtype: string
+  - name: language
+    dtype: string
+  splits:
+  - name: train
+    num_examples: 491
+  config_name: default
+---
+
+# tetttttt
+
+This is a merged speech dataset containing 491 audio segments from 2 source datasets.
+
+## Dataset Information
+
+- **Total Segments**: 491
+- **Speakers**: 2
+- **Languages**: tr
+- **Emotions**: angry, neutral, happy
+- **Original Datasets**: 2
+
+## Dataset Structure
+
+Each example contains:
+- `audio`: Audio file (WAV format, original sampling rate preserved)
+- `text`: Transcription of the audio
+- `speaker_id`: Unique speaker identifier (made unique across all merged datasets)
+- `emotion`: Detected emotion (neutral, happy, sad, etc.)
+- `language`: Language code (en, es, fr, etc.)
+
+## Usage
+
+### Loading the Dataset
+
+```python
+from datasets import load_dataset
+
+# Load the dataset
+dataset = load_dataset("Codyfederer/tetttttt")
+
+# Access the training split
+train_data = dataset["train"]
+
+# Example: Get first sample
+sample = train_data[0]
+print(f"Text: {sample['text']}")
+print(f"Speaker: {sample['speaker_id']}")
+print(f"Language: {sample['language']}")
+print(f"Emotion: {sample['emotion']}")
+
+# Play audio (requires audio libraries)
+# sample['audio']['array'] contains the audio data
+# sample['audio']['sampling_rate'] contains the sampling rate
+```
+
+### Alternative: Load from JSONL
+
+```python
+from datasets import Dataset, Audio, Features, Value
+import json
+
+# Load the JSONL file
+rows = []
+with open("data.jsonl", "r", encoding="utf-8") as f:
+    for line in f:
+        rows.append(json.loads(line))
+
+features = Features({
+    "audio": Audio(sampling_rate=None),
+    "text": Value("string"),
+    "speaker_id": Value("string"),
+    "emotion": Value("string"),
+    "language": Value("string")
+})
+
+dataset = Dataset.from_list(rows, features=features)
+```
+
+### Dataset Structure
+
+The dataset includes:
+- `data.jsonl` - Main dataset file with all columns (JSON Lines)
+- `*.wav` - Audio files under `audio_XXX/` subdirectories  
+- `load_dataset.txt` - Python script for loading the dataset (rename to .py to use)
+
+JSONL keys:
+- `audio`: Relative audio path (e.g., `audio_000/segment_000000_speaker_0.wav`)
+- `text`: Transcription of the audio
+- `speaker_id`: Unique speaker identifier
+- `emotion`: Detected emotion
+- `language`: Language code
+
+## Speaker ID Mapping
+
+Speaker IDs have been made unique across all merged datasets to avoid conflicts.
+For example:
+- Original Dataset A: `speaker_0`, `speaker_1`
+- Original Dataset B: `speaker_0`, `speaker_1`
+- Merged Dataset: `speaker_0`, `speaker_1`, `speaker_2`, `speaker_3`
+
+Original dataset information is preserved in the metadata for reference.
+
+## Data Quality
+
+This dataset was created using the Vyvo Dataset Builder with:
+- Automatic transcription and diarization
+- Quality filtering for audio segments
+- Music and noise filtering
+- Emotion detection
+- Language identification
+
+## License
+
+This dataset is released under the Creative Commons Attribution 4.0 International License (CC BY 4.0).
+
+## Citation
+
+```bibtex
+@dataset{vyvo_merged_dataset,
+  title={tetttttt},
+  author={Vyvo Dataset Builder},
+  year={2025},
+  url={https://huggingface.co/datasets/Codyfederer/tetttttt}
+}
+```
+
+This dataset was created using the Vyvo Dataset Builder tool.

dataset_info.json

@@ -0,0 +1,49 @@
+{
+  "dataset_name": "tetttttt",
+  "description": "Merged speech dataset containing 491 segments from 2 source datasets",
+  "features": {
+    "audio": {
+      "_type": "Audio",
+      "sampling_rate": null
+    },
+    "text": {
+      "_type": "Value",
+      "dtype": "string"
+    },
+    "speaker_id": {
+      "_type": "Value",
+      "dtype": "string"
+    },
+    "emotion": {
+      "_type": "Value",
+      "dtype": "string"
+    },
+    "language": {
+      "_type": "Value",
+      "dtype": "string"
+    }
+  },
+  "splits": {
+    "train": {
+      "name": "train",
+      "num_examples": 491
+    }
+  },
+  "total_segments": 491,
+  "speakers": [
+    "speaker_0",
+    "speaker_1"
+  ],
+  "emotions": [
+    "angry",
+    "neutral",
+    "happy"
+  ],
+  "languages": [
+    "tr"
+  ],
+  "original_datasets": [
+    "1DPlg5pXztI",
+    "-WOKqeQsYzs"
+  ]
+}

load_dataset.txt

@@ -0,0 +1,46 @@
+# Dataset Loading Script
+# Save this as load_dataset.py to use
+
+import csv
+import os
+from datasets import Dataset, Audio, Value, Features
+
+def load_dataset():
+    # Define features
+    features = Features({
+        # Preserve original sampling rates by not forcing a fixed rate
+        "audio": Audio(sampling_rate=None),
+        "text": Value("string"),
+        "speaker_id": Value("string"),
+        "emotion": Value("string"),
+        "language": Value("string")
+    })
+    
+    # Load data from CSV
+    data = {
+        "audio": [],
+        "text": [],
+        "speaker_id": [],
+        "emotion": [],
+        "language": []
+    }
+    
+    # Read JSONL
+    import json
+    with open("data.jsonl", "r", encoding="utf-8") as f:
+        for line in f:
+            obj = json.loads(line)
+            data["audio"].append(obj["audio"])  # relative path within repo
+            data["text"].append(obj.get("text", ""))
+            data["speaker_id"].append(obj.get("speaker_id", ""))
+            data["emotion"].append(obj.get("emotion", "neutral"))
+            data["language"].append(obj.get("language", "en"))
+    
+    # Create dataset
+    dataset = Dataset.from_dict(data, features=features)
+    return dataset
+
+# For direct loading
+if __name__ == "__main__":
+    dataset = load_dataset()
+    print(f"Dataset loaded with {len(dataset)} examples")

train-00000-of-00001.parquet

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:fe002e0376c3e430421ed1521dcb70e177d323f4c0119936f141a03e78741c23
+size 229484945