docs: add INTERSPEECH 2026 paper banner, badges, and citation
Browse files
README.md
CHANGED
|
@@ -11,6 +11,16 @@ tags:
|
|
| 11 |
|
| 12 |
# A Data-Centric Framework for Addressing Phonetic and Prosodic Challenges in Russian Speech Generative Models
|
| 13 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 14 |
Russian speech synthesis presents distinctive challenges, including vowel reduction, consonant devoicing, variable stress patterns, homograph ambiguity, and unnatural intonation. This paper introduces Balalaika, a novel dataset comprising more than 2,000 hours of studio-quality Russian speech with comprehensive textual annotations, including punctuation and stress markings. Experimental results show that models trained on Balalaika significantly outperform those trained on existing datasets in both speech synthesis and enhancement tasks.
|
| 15 |
|
| 16 |
Paper: [A Data-Centric Framework for Addressing Phonetic and Prosodic Challenges in Russian Speech Generative Models](https://huggingface.co/papers/2507.13563)
|
|
@@ -344,3 +354,18 @@ Comply with each component’s original license in addition to the above:
|
|
| 344 |
|
| 345 |
- Email: kborodin.research@gmail.com
|
| 346 |
- Telegram: [@korallll_ai](https://t.me/korallll_ai)
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 11 |
|
| 12 |
# A Data-Centric Framework for Addressing Phonetic and Prosodic Challenges in Russian Speech Generative Models
|
| 13 |
|
| 14 |
+
[](https://arxiv.org/abs/2507.13563)
|
| 15 |
+
[](https://arxiv.org/abs/2507.13563)
|
| 16 |
+
[](https://github.com/lab260ru/balalaika)
|
| 17 |
+
|
| 18 |
+
> [!IMPORTANT]
|
| 19 |
+
> **Official dataset for our INTERSPEECH 2026 paper**
|
| 20 |
+
> *"A Data-Centric Framework for Addressing Phonetic and Prosodic Challenges in Russian Speech Generative Models"* ([arXiv:2507.13563](https://arxiv.org/abs/2507.13563)).
|
| 21 |
+
> Part of the **Balalaika** Russian speech data-processing pipeline — code: [https://github.com/lab260ru/balalaika](https://github.com/lab260ru/balalaika).
|
| 22 |
+
> If you use this resource, please [cite it](#citation).
|
| 23 |
+
|
| 24 |
Russian speech synthesis presents distinctive challenges, including vowel reduction, consonant devoicing, variable stress patterns, homograph ambiguity, and unnatural intonation. This paper introduces Balalaika, a novel dataset comprising more than 2,000 hours of studio-quality Russian speech with comprehensive textual annotations, including punctuation and stress markings. Experimental results show that models trained on Balalaika significantly outperform those trained on existing datasets in both speech synthesis and enhancement tasks.
|
| 25 |
|
| 26 |
Paper: [A Data-Centric Framework for Addressing Phonetic and Prosodic Challenges in Russian Speech Generative Models](https://huggingface.co/papers/2507.13563)
|
|
|
|
| 354 |
|
| 355 |
- Email: kborodin.research@gmail.com
|
| 356 |
- Telegram: [@korallll_ai](https://t.me/korallll_ai)
|
| 357 |
+
|
| 358 |
+
## Citation
|
| 359 |
+
|
| 360 |
+
If you use this resource, please cite our INTERSPEECH 2026 paper:
|
| 361 |
+
|
| 362 |
+
```bibtex
|
| 363 |
+
@inproceedings{borodin2026balalaika,
|
| 364 |
+
title = {A Data-Centric Framework for Addressing Phonetic and Prosodic Challenges in Russian Speech Generative Models},
|
| 365 |
+
author = {Borodin, Kirill and Vasiliev, Nikita and Kudryavtsev, Vasiliy and Maslov, Maxim and Gorodnichev, Mikhail and Rogov, Oleg and Mkrtchian, Grach},
|
| 366 |
+
booktitle = {Proc. INTERSPEECH 2026},
|
| 367 |
+
year = {2026},
|
| 368 |
+
note = {arXiv:2507.13563},
|
| 369 |
+
url = {https://arxiv.org/abs/2507.13563}
|
| 370 |
+
}
|
| 371 |
+
```
|