README.md

metadata

dataset_info:
  features:
    - name: audio
      dtype: audio
    - name: text
      dtype: string
    - name: source
      dtype: string
    - name: sample_rate
      dtype: int64
    - name: speaker
      dtype: string
  splits:
    - name: train
      num_bytes: 2942869578
      num_examples: 13000
  download_size: 3303296950
  dataset_size: 2942869578
configs:
  - config_name: default
    data_files:
      - split: train
        path: data/train-*
license: cc
task_categories:
  - text-to-speech
  - automatic-speech-recognition
language:
  - tr
size_categories:
  - 10K<n<100K

Synthetic Turkish TTS Data

This dataset was created by generating synthetic Turkish text across multiple speech scenarios. The text was produced in the following domains: finance_master, cs_master, parcel_delivery, ecommerce, telecom, isp_support, technical_support, subscription, insurance, health_appointments, public_services, education_registration, and daily_speech.

These synthetic texts were then synthesized with a high-quality Turkish TTS model. The dataset is intended to be used as synthetic training data for future Turkish TTS model training.

Dataset Summary

• Total number of samples: 13,000
• Total number of speakers: 4
• Total duration: 28 hours 50 minutes 13 seconds

Speaker Distribution

• ali: 6 hours 50 minutes 57 seconds (3250 wav files)
• zeynep: 6 hours 50 minutes 35 seconds (3250 wav files)
• leyla: 7 hours 10 minutes 55 seconds (3250 wav files)
• alev: 7 hours 57 minutes 46 seconds (3250 wav files)

Columns

• audio: audio file
• text: corresponding transcription
• source: scenario/domain from which the sample was generated
• sample_rate: audio sampling rate
• speaker: speaker name

License

This dataset is released under the CC BY 4.0 license. You may use, share, and adapt the dataset provided that proper attribution is given.

Audio Generation

The audio samples in this dataset were generated using the Freya AI API.

Reference: https://www.linkedin.com/company/107923818/