Dataset Viewer
Duplicate
The dataset viewer is not available for this split.
Server error while post-processing the rows. Please report the issue.
Error code:   RowsPostProcessingError

Need help to make the dataset viewer work? Make sure to review how to configure the dataset viewer, and open a discussion for direct support.

Not: Bu veri setinin dokümantasyonu Türk yapay zeka topluluğuna katkı sağlamak amacıyla VeriPazarı tarafından Türkçeye çevrilmiştir. Orijinal veri seti Codyfederer tarafından geliştirilmiş olup, VeriPazarı tarafından Türk AI ekosistemi için arşivlenmiştir.

🔗 Orijinal Kaynak: Codyfederer/tr-full-dataset 🔗 Derleyen Platform: VeriPazarı

TR-Full_dataset (Duygu Etiketli Türkçe Ses Veri Seti)

Bu veri seti, 88 kaynak veri setinden derlenmiş 41.427 ses segmentini (parçasını) içeren birleştirilmiş (merged) bir ses veri setidir.

Veri Seti Bilgileri

  • Toplam Segment: 41.427
  • Konuşmacılar (Speakers): 222
  • Diller: tr (Türkçe)
  • Duygular (Emotions): neutral (nötr), angry (kızgın), sad (üzgün), happy (mutlu)
  • Orijinal Veri Setleri: 88 adet farklı kaynak

Veri Seti Yapısı

Her bir örnek şunları içerir:

  • audio: Ses dosyası (WAV formatında, orijinal örnekleme hızı korunmuştur)
  • text: Sesin transkripsiyonu (metin hali)
  • speaker_id: Benzersiz konuşmacı kimliği (birleştirilen tüm veri setleri arasında benzersiz hale getirilmiştir)
  • emotion: Tespit edilen duygu (nötr, mutlu, üzgün vb.)
  • language: Dil kodu (tr, en, es, fr vb.)

Kullanım

Veri Setini Yükleme (Hugging Face Datasets ile)

from datasets import load_dataset

# Veri setini yükle
dataset = load_dataset("Codyfederer/tr-full-dataset")

# Eğitim (train) ayrımına erişim sağla
train_data = dataset["train"]

# Örnek: İlk veriyi al
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']}")

# Sesi oynatma (ses kütüphaneleri gerektirir)
# sample['audio']['array'] ses verisini içerir
# sample['audio']['sampling_rate'] örnekleme hızını içerir

Alternatif: JSONL Dosyasından Yükleme

from datasets import Dataset, Audio, Features, Value
import json

# JSONL dosyasını yükle
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)

Veri Seti Dosya Yapısı

Veri seti şunları içerir:

  • data.jsonl - Tüm sütunları içeren ana veri seti dosyası (JSON Lines)
  • *.wav - audio_XXX/ alt dizinlerindeki ses dosyaları
  • load_dataset.txt - Veri setini yüklemek için Python betiği (kullanmak için adını .py olarak değiştirin)

JSONL Anahtarları (Keys):

  • audio: Göreceli ses yolu (örn: audio_000/segment_000000_speaker_0.wav)
  • text: Sesin transkripsiyonu
  • speaker_id: Benzersiz konuşmacı kimliği
  • emotion: Tespit edilen duygu
  • language: Dil kodu

Konuşmacı ID (Speaker ID) Eşleştirmesi

Çakışmaları önlemek adına Konuşmacı ID'leri, birleştirilen tüm veri setleri genelinde benzersiz hale getirilmiştir. Örneğin:

  • Orijinal Veri Seti A: speaker_0, speaker_1
  • Orijinal Veri Seti B: speaker_0, speaker_1
  • Birleştirilmiş (Merged) Veri Seti: speaker_0, speaker_1, speaker_2, speaker_3

Orijinal veri seti bilgileri, referans olması amacıyla meta verilerde (metadata) korunmaktadır.

Veri Kalitesi

Bu veri seti, Vyvo Dataset Builder kullanılarak aşağıdaki özelliklerle oluşturulmuştur:

  • Otomatik transkripsiyon ve konuşmacı ayrıştırma (diarization)
  • Ses segmentleri için kalite filtrelemesi
  • Müzik ve gürültü filtrelemesi
  • Duygu tespiti (Emotion detection)
  • Dil tanıma (Language identification)

Lisans

Bu veri seti Creative Commons Attribution 4.0 International License (CC BY 4.0) altında yayınlanmıştır.

Atıf (Citation)

@dataset{vyvo_merged_dataset,
  title={TR-Full_dataset},
  author={Vyvo Dataset Builder},
  year={2025},
  url={https://huggingface.co/datasets/Codyfederer/tr-full-dataset}
}

Bu veri seti Vyvo Dataset Builder aracı kullanılarak oluşturulmuştur.


Bu dosya veripazari.com.tr topluluğu tarafından Hugging Face altyapısında barındırılmaktadır. Orijinal emeğe saygı kuralımız gereği lisans ve model isimleri korunmuştur.

Downloads last month
5,161