FTV and Microsoft Partner to Launch AI News Anchor "Min-Hsi"

Generative AI technology is revolutionizing traditional AI usage and fostering diverse innovative applications across various fields, with the media industry actively participating in this wave.

Microsoft Azure AI Speech's TTS (see Note 1 below) and Viseme technology (see Note 2 below) are the keys to achieving anchor Min-Hsi's natural and smooth human voice. This technology differs from traditional mechanical text-to-speech services because it employs rich models and offers multiple options, giving it an advantage in various applications.

Azure AI Speech also provides support for custom models, meaning parameters such as voice style, speed, and pitch can be adjusted based on specific needs. This flexibility allows for the generation of speech that is closer to a real human anchor, incorporating elements like emotional expression and intonation, which further enhances the natural feel and emotional connection of the speech synthesis. This is highly valuable for creating personalized speech synthesis applications and meeting the needs of different scenarios and goals.

The database includes multiple languages such as Chinese, English, Taiwanese, and Hakka. Ongoing training is conducted to develop brand-new image models capable of presenting finer facial muscle textures. The goal of this effort is to ensure the virtual anchor's overall appearance, voice, and body language are more realistic and natural.

After the AI anchor went live in June, it immediately sparked an enthusiastic response from the market and the audience. This not only deepened the audience's sense of connection with the virtual anchor but also opened up brand-new possibilities for the media industry, further highlighting the importance of virtual anchors in modern media and communication. This success story sets a precedent for the application of virtual anchor technology and media innovation.

Notes:

• TTS Technology: TTS stands for Text-to-Speech, which allows text to be converted into natural, fluent human voices. Microsoft Azure AI Speech uses pre-built neural voice models capable of mimicking over 400 human-like pronunciation styles, providing exceptionally high voice quality.

• Viseme Technology: A Viseme is a technology based on the shape of the mouth during pronunciation, which helps align speech more naturally with mouth movements. This technology allows the mouth shapes of animations and virtual characters to match actual speech more closely, providing a more vivid experience.