toplogo
Войти

FeruzaSpeech: A 60-Hour High-Quality Uzbek Speech Corpus with Dual Alphabet Transcripts and Context


Основные понятия
FeruzaSpeech is a high-quality Uzbek speech corpus that provides 60 hours of recordings from a single native female speaker, with transcripts in both Cyrillic and Latin alphabets, to support the development of speech recognition and synthesis technologies for the Uzbek language.
Аннотация

The FeruzaSpeech corpus is a significant contribution to the field of Uzbek speech technology. It provides 60 hours of high-quality, single-channel, 16-bit audio recordings at 16kHz, with transcripts in both the Cyrillic and Latin alphabets. The corpus consists of excerpts from a classic Uzbek novel, "Choliqushi," and BBC Uzbek news articles, read by a native female speaker from Tashkent, Uzbekistan.

The dataset is divided into training, development, and test sets, with the training set including both the book and news excerpts, and the development and test sets containing only the news excerpts. The audio segments are longer than those in other Uzbek speech corpora, with an average length of 16.39 seconds, containing one to two full sentences.

Experiments conducted using the FeruzaSpeech corpus, in combination with the existing Common Voice Uzbek Dataset and Uzbek Speech Corpus, have shown significant improvements in word error rates (WERs) for automatic speech recognition (ASR) models. The best WER on the Common Voice test set was 11.17%, the best WER on the FeruzaSpeech test set was 4.05%, and the best WER on the Uzbek Speech Corpus test set was 11.67%.

The availability of dual-alphabet transcripts in FeruzaSpeech is a unique feature, as it addresses the challenge of accurately converting between the Cyrillic and Latin alphabets used in Uzbekistan. This corpus complements the existing Uzbek speech datasets and is expected to contribute to the advancement of speech recognition and synthesis technologies for the Uzbek language.

edit_icon

Настроить сводку

edit_icon

Переписать с помощью ИИ

edit_icon

Создать цитаты

translate_icon

Перевести источник

visual_icon

Создать интеллект-карту

visit_icon

Перейти к источнику

Статистика
Biometric passports, which are currently in use, will become almost useless from January 1, 2019. On July 20, 562 cases of the disease were recorded in Uzbekistan.
Цитаты
"— Don't say bread! — he said. — Don't utter the word bread!" "after that he looked like if you don't tell I will"

Дополнительные вопросы

How can the FeruzaSpeech corpus be further expanded to include more diverse speakers and content, while maintaining the high-quality and consistent voice characteristics?

To expand the FeruzaSpeech corpus while ensuring high quality and consistent voice characteristics, several strategies can be employed. First, the inclusion of additional native speakers from various regions of Uzbekistan can enhance the diversity of accents and dialects within the corpus. This can be achieved by conducting targeted recording sessions with speakers from different backgrounds, ensuring that they are trained to deliver the content in a manner similar to the original speaker, Feruza. Second, the content can be diversified by incorporating a wider range of genres beyond audiobooks and news, such as poetry, folklore, and conversational dialogues. This would not only enrich the corpus but also provide a broader context for speech recognition and synthesis applications. To maintain voice consistency, it is crucial to establish clear guidelines for recording sessions, including vocal delivery style, pacing, and emotional tone. Utilizing a voice coach or a trained director during the recording process can help achieve this uniformity. Additionally, employing advanced audio processing techniques can ensure that the recordings are of high quality, with minimal background noise and optimal clarity. Finally, regular updates and expansions of the corpus can be facilitated by creating a community of contributors who can provide feedback and suggest new content, thereby fostering a collaborative approach to corpus development.

What are the potential challenges in developing accurate Cyrillic-to-Latin and Latin-to-Cyrillic conversion tools for the Uzbek language, and how can they be addressed?

Developing accurate Cyrillic-to-Latin and Latin-to-Cyrillic conversion tools for the Uzbek language presents several challenges. One significant issue is the presence of phonetic nuances and specific characters in the Uzbek language that do not have direct equivalents in the other script. For instance, the soft sign (ь) in Cyrillic can be problematic, as it may be lost or misrepresented during conversion, leading to inaccuracies in pronunciation and meaning. To address these challenges, it is essential to create a comprehensive mapping system that accounts for all phonetic variations and contextual uses of characters in both scripts. This could involve linguistic research to understand the specific phonetic characteristics of Uzbek and how they are represented in each alphabet. Additionally, machine learning techniques can be employed to train conversion algorithms on large datasets that include both scripts. By using a diverse range of text samples, the algorithms can learn to recognize patterns and make more accurate conversions. Continuous testing and refinement of these tools, based on user feedback and real-world applications, will also be crucial in improving their accuracy and reliability.

How can the FeruzaSpeech corpus be leveraged to explore the intersection between speech technology and cultural preservation, particularly in the context of language transitions and script changes in Uzbekistan?

The FeruzaSpeech corpus can serve as a vital resource for exploring the intersection of speech technology and cultural preservation, especially in light of Uzbekistan's transition from Cyrillic to Latin script. By providing a dual-alphabet corpus, FeruzaSpeech not only facilitates research in automatic speech recognition (ASR) and text-to-speech (TTS) technologies but also plays a crucial role in documenting and preserving the linguistic heritage of the Uzbek language. One way to leverage this corpus is by developing educational tools and applications that help speakers transition between scripts. For instance, interactive language learning platforms can utilize the corpus to teach users how to read and pronounce words in both Cyrillic and Latin scripts, thereby promoting literacy and understanding of the language's evolution. Furthermore, the corpus can be used to create culturally relevant content that reflects the rich history and traditions of Uzbekistan. By integrating folklore, poetry, and historical narratives into speech technology applications, developers can foster a deeper connection between users and their cultural heritage. Additionally, researchers can analyze the corpus to study the impact of script changes on language usage, dialect variation, and identity among Uzbek speakers. This research can inform policymakers and educators about the importance of preserving linguistic diversity and ensuring that language transitions do not lead to the erosion of cultural identity. In summary, the FeruzaSpeech corpus is not only a tool for advancing speech technology but also a means of preserving and promoting the cultural richness of the Uzbek language during a significant period of transition.
0
star