---
license: mit
task_categories:
- text-to-speech
language:
- en
tags:
- mean-opinion-score
- text-to-speech
- speech
- lj-speech
- speech-quality
- intelligibility
- naturalness
- human-evaluation
- mturk
- evaluation
- crowdsourcing
pretty_name: LJ Speech MOS 120 Dataset
size_categories:
- 10K<n<100K
---
# LJS-MOS-120: Human MOS Ratings for 120 Samples of the LJ Speech Dataset

**LJS-MOS-120** provides Mean Opinion Score (MOS) ratings for 120 text-to-speech (TTS) samples based on the [LJ Speech](https://keithito.com/LJ-Speech-Dataset) dataset. Each sample was rated by human annotators for **intelligibility** and **naturalness** across four experimental TTS conditions. The dataset follows the **Tidy data format**, with one row per rating per dimension.

Ratings were collected via **Amazon Mechanical Turk (MTurk)** in 2022. Each entry includes the worker ID (anonymized), gender (`male`, `female`), age group (`18–29`, `30–49`, `50+`), and headphone usage.

## TTS Conditions

- `gt`: Ground truth (original human recording)
- `wg`: Ground truth Mel-spectrogram resynthesized with WaveGlow
- `impl`: Mel-spectrogram generated using an **implicit duration model** ([Zenodo](https://zenodo.org/records/10200955)) and synthesized with WaveGlow
- `expl`: Mel-spectrogram generated using an **explicit duration model** ([Zenodo](https://zenodo.org/records/10107104)) and synthesized with WaveGlow

## Dataset Format

Each row contains:

- `audio`: path to the WAV file
- `file`: filename (e.g., `LJ002-0092`)
- `condition`: one of `gt`, `wg`, `impl`, `expl`
- `score_dimension`: `intelligibility` or `naturalness`
- `score`: integer rating (1–5)
- `worker_id`: anonymized integer ID
- `gender`, `age_group`, `used_headphone`: rater metadata
- `assignment_id`, `hit_id`, `state`: MTurk metadata

## License

- LJ Speech: Public Domain (CC0)
- This dataset: MIT License

## Acknowledgements

- Funded by the Deutsche Forschungsgemeinschaft (DFG, German Research Foundation) – Project-ID 416228727 – CRC 1410