What is Spark Character
Spark Character is an independent model (character simulation API) launched by iFLYTEK, designed specifically for emotional interaction. It offers control options such as character settings, plot performance, and language style, and supports model fine-tuning. In double-blind experience tests with C-end users at scale, Spark Character's performance is already leading in the industry. By calling Spark Character, agents can achieve both "IQ and EQ" online, providing a more humanized and emotionally rich experience in interactions with users.
Main features of Spark Character
- Character settings
- Customize virtual character traits: Users can customize the virtual character's worldview, social identity, background experience, personality traits, and interests to create a distinctive virtual personality. The richer and more detailed the character definition, the more three-dimensional and full the interactive experience presented.
- Form a personalized AI persona: Form an initial persona based on personality simulation, emotion understanding, and expression style, then combine specific knowledge learning and dialogue memory learning to form a more personalized AI persona.
- Plot performance
- Proactively initiate conversations: The character can not only passively respond to the user's speech but also simulate proactively initiating conversations. Developers only need to design the trigger timing for proactive conversations, allowing users to experience unexpected interactive surprises.
- Respond in context: Accurately understand user semantics and generate natural responses that highly fit the set scenario and character settings, making conversations more realistic and natural.
- Language style control
- Diverse language styles: Supports control options for multiple language styles, such as formal, humorous, and friendly, to adapt to different scenarios and user needs, making interactions more aligned with user expectations and preferences.
- Controllable voice expression: In voice interaction, it can achieve controllable voice expression, making the voice's tone, speed, timbre, etc. more consistent with character settings and emotional expression.
- Model fine-tuning: Supports model fine-tuning. Users can further optimize and adjust the Spark Character model according to specific application scenarios and needs to improve its performance and adaptability in specific fields.
- Memory and reasoning capabilities
- Maintain conversation memory: Records the conversation history between the character and the user, supports maintaining contextual memory across multiple conversation segments, and ensures consistency and coherence of interactions. Even if the user temporarily interrupts, the character can seamlessly pick up the previous memory when visiting again, making the experience more immersive and smooth.
- Possess reasoning ability: Integrates powerful reasoning ability, can execute instructions based on user intent, and is more natural in processing multiple types of data such as text, images, and voice, and in achieving interaction.
How to use Spark Character
- Register and log in:Go to iFLYTEK's official website, click to apply, register, and log in. After binding a mobile phone number, enter the experience application form, fill in the relevant information, and submit the application. It can be used after approval.
- Create an application: Log in toiFLYTEK Open Platform, click to enter Console - My Applications, and create an application.
- Activate the service: After the application is created, click the application name, then click the left navigation to switch to "Spark Cognitive Large Model" - "Role Simulation API", and click "Buy Now" to claim the trial package quota.
- Obtain identity credentials: The Role Simulation API performs authentication and authorization via AK/SK. You can obtain the application's API key and API secret in the console under "Spark Cognitive Large Model" - "Role Simulation API".
- Customize persona
- Define character traits: As needed, set the worldview, social identity, background experience, personality traits, and interests for the virtual character.
- Form a personalized persona: Form an initial persona based on personality simulation, emotion understanding, and expression style, then combine specific knowledge learning and dialogue memory learning to form a more personalized AI persona.
- Call the API
- After integrating role simulation into a third-party system, interact with the Spark persona by calling the API.
- Users enter their own remarks, and the Spark persona service instantly generates responses that fit the scenario and character settings.
- View service usage information: In the console under "Spark Cognitive Large Model" - "Role Simulation API", view the consumed word count and the remaining available quota.
Application scenarios of Spark persona
- Medical consultation: Spark persona can simulate the role of a doctor, provide users with some basic medical consultation and advice, and help users understand their own health conditions.
- Health management: In health management, Spark persona can remind users to take medication on time and have regular physical examinations, and provide personalized health management advice based on the user's health data.
- Virtual character interaction: Users can create or select different virtual characters to interact with the Spark persona and enjoy an immersive entertainment experience.
- Social chat: Spark persona can serve as a virtual social partner, chat and communicate with users, and provide companionship and the fun of social interaction.
- Customer service: In enterprise customer service, Spark persona can simulate the role of customer service personnel, provide users with professional consultation services, and answer users' doubts and questions.
- Product introduction: Spark persona can be used for product introduction and promotion, and by simulating the role of a salesperson, introduce product functions and features to users in detail.