Upload README.md with huggingface_hub

README.md

@@ -1,51 +1,38 @@
-# KazEmoTTS: A Kazakh Emotional Text-to-Speech Dataset
+# KazEmoTTS: A Multi-Speaker Emotional Kazakh Text-to-Speech Dataset
 
-**Dataset Summary:**
+**Repository:** [Insert GitHub Repository Link Here -  replace with actual link once available]
 
-The KazEmoTTS dataset is a collection of 54,760 audio-text pairs totaling 74.85 hours of emotional speech in the Kazakh language.  The dataset includes recordings from three narrators (one female, two male) expressing six emotions: neutral, angry, happy, sad, scared, and surprised.  A text-to-speech (TTS) model trained on this dataset is also available.  The dataset aims to facilitate research and development in emotional TTS for the Kazakh language.
 
+**Summary:**
 
-**Dataset Characteristics:**
+The KazEmoTTS dataset is a new resource for emotional Kazakh text-to-speech (TTS).  It contains 54,760 audio-text pairs totaling 74.85 hours of speech, encompassing six emotions (neutral, angry, happy, sad, scared, and surprised). The data includes recordings from one female and two male narrators.  A TTS model trained on this dataset achieved Mean Character Distance (MCD) scores between 6.02 and 7.67 and Mean Opinion Scores (MOS) between 3.51 and 3.57, as measured through objective and subjective evaluations. The dataset and trained model are publicly available to promote further research and reproducibility.
 
-* **Size:** 54,760 audio-text pairs
-* **Duration:** 74.85 hours (34.23 hours female, 40.62 hours male)
-* **Languages:** Kazakh
-* **Emotions:** Neutral, Angry, Happy, Sad, Scared, Surprised
-* **Number of Speakers:** 3 (1 female, 2 male)
 
+**Abstract:**
 
-**Data Fields:**
+This study presents the KazEmoTTS dataset, a large-scale resource for building emotional Kazakh TTS systems.  The dataset comprises a substantial amount of speech data (54,760 audio-text pairs, 74.85 hours total duration) with diverse emotional expressions (neutral, angry, happy, sad, scared, surprised) spoken by three narrators (one female, two male).  Evaluation of a TTS model trained on KazEmoTTS reveals promising performance metrics (MCD: 6.02-7.67; MOS: 3.51-3.57), demonstrating the dataset's suitability for developing high-quality emotional TTS applications.  The dataset's open-access nature is intended to facilitate further research and development in the field.
 
-The dataset contains audio files and corresponding text transcripts.  Specific field names and formats are available in the linked repository.
 
+**Dataset Statistics:**
 
-**Evaluation Metrics:**
-
-The quality of the synthesized speech generated by a model trained on the KazEmoTTS dataset was evaluated using Mean Opinion Score (MOS) and Mel-Cepstral Distortion (MCD).  MOS ranged from 3.51 to 3.57, and MCD ranged from 6.02 to 7.67.
-
-
-**License:**
-
-[Insert License Here -  This information is missing from the provided text]
-
-
-**Citation:**
+| Statistic             | Value          |
+|----------------------|-----------------|
+| Number of Audio Files | 54,760         |
+| Total Duration        | 74.85 hours     |
+| Female Narrator Duration | 34.23 hours     |
+| Male Narrator Duration  | 40.62 hours     |
+| Number of Emotions    | 6               |
+| Emotions              | Neutral, Angry, Happy, Sad, Scared, Surprised |
 
-[Insert Citation Here - This information is missing from the provided text]
 
+**Evaluation Metrics:**
 
-**Contact:**
-
-[Insert Contact Information Here - This information is missing from the provided text]
-
-
-**How to Use:**
-
-Instructions on how to access and utilize the dataset are provided within the linked repository.
-
+| Metric | Range      |
+|--------|------------|
+| MCD    | 6.02 - 7.67 |
+| MOS    | 3.51 - 3.57 |
 
-**Links:**
 
-* **Dataset Repository:** [Insert GitHub Repository Link Here]
-* **Model Repository:** [Insert GitHub Model Repository Link Here, if different from dataset]
+**Citations:**
 
+*(No citations provided in the original text.  Please add citations here if available from the original source.)*