We often use natural language dialogue systems in our daily life, such as Apple’s Siri, Microsoft’s XiaoIce and Amazon’s Alexa. Most of the current dialogue systems can only do a single round of dialogue, which can help users complete some simple tasks, such as asking the weather, checking stocks (if doing multiple rounds of dialogue, also add some simple processing on the basis of a single round of dialogue).

Dialogue systems have developed very rapidly in recent years, especially in the number of papers at the top NLP conferences. If natural language processing is the jewel in the crown of AI, then the dialogue system is the “jewel in the crown of NLP”. Task-based dialogues represented by Apple’s Siri and non-task-based (chat-based) dialogues represented by Microsoft’s XiaoIce are particularly concerned by academia and the industry.

The intelligent dialogue system is to enable the machine to understand the intention of human language and perform specific tasks or answer through effective human-computer interaction under the support of various intelligent algorithms. With the continuous development of technology, task-based dialogue systems have been widely used in virtual personal assistants, smart homes, smart cars (vehicle voice) and other fields. Chatting dialogue systems have also found application scenarios in the fields of entertainment and emotional escort.

However, we should see that these traditional dialogue systems have some problems, such as inaccurate semantic understanding resulting in incorrect answers, inconsistent identities and personalities displayed in dialogues, which make it difficult to gain user trust, and possible moral and ethical risks in dialogue interactions. Therefore, how to avoid and solve these problems and develop next-generation dialogue systems with better interaction effects has gradually become a hot research topic in the industry.

In order to facilitate the development and application of smarter and more user-friendly voice assistants, Nexdata has 40,000 hours conversational speech data, covering single and multi-person conversations, multiple languages and various scenerios. These ready-to-go datasets have legalized copyright and gurantee 95% accuracy rate of sentence.

Mandarin Conversational Speech Data

1950 speakers participated in the recording, and conducted face-to-face communication in a natural way. They had free discussion on a number of given topics, with a wide range of fields; the voice was natural and fluent, in line with the actual dialogue scene. Text is transferred manually, with high accuracy.

Cantonese Conversational Speech Data

995 local Cantonese speakers participated in the recording, and conducted face-to-face communication in a natural way. They had free discussion on a number of given topics, with a wide range of fields; the voice was natural and fluent, in line with the actual dialogue scene. Text is transferred manually, with high accuracy.

Hindi Conversational Speech Data

About 1,000 speakers participated in the recording, and conducted face-to-face communication in a natural way. They had free discussion on a number of given topics, with a wide range of fields; the voice was natural and fluent, in line with the actual dialogue scene. Text is transferred manually, with high accuracy.

Italian Conversational Speech Data

About 700 speakers participated in the recording, and conducted face-to-face communication in a natural way. They had free discussion on a number of given topics, with a wide range of fields; the voice was natural and fluent, in line with the actual dialogue scene. Text is transferred manually, with high accuracy.

French Conversational Speech Data

About 700 speakers participated in the recording, and conducted face-to-face communication in a natural way. They had free discussion on a number of given topics, with a wide range of fields; the voice was natural and fluent, in line with the actual dialogue scene. Text is transferred manually, with high accuracy.

End

If you want to know more details about the datasets or how to acquire, please feel free to contact us: info@nexdata.ai.

The future intelligent system will increasingly rely on high-quality datasets to optimize decision-making and automated processes. In the era of data, companies and researchers need to continuously improve their ability of data collection and annotation to make sure the efficiency and accuracy of AI models. To gain an advantageous position in fiercely competitive market, we must laid a solid foundation in data.