ComfyUI Node: IndexTTS2 Emotion Vector

IndexTTS2 Emotion Vector:

The IndexTTS2EmotionVector node is designed to facilitate the integration of emotional nuances into text-to-speech (TTS) systems by utilizing emotion vectors. This node allows you to define and manipulate the emotional tone of synthesized speech, enhancing the expressiveness and realism of AI-generated voices. By converting predefined emotional states into a vector format, the node provides a structured approach to influence the emotional output of TTS systems. This capability is particularly beneficial for applications requiring nuanced emotional expression, such as virtual assistants, storytelling, and interactive media. The node ensures that the emotional intensity is balanced and within acceptable limits, preventing overly exaggerated expressions that could detract from the intended communication.

IndexTTS2 Emotion Vector Input Parameters:

happy

This parameter represents the intensity of the "happy" emotion in the vector. It influences how cheerful or joyful the synthesized speech will sound. The value must be a non-negative float, with a typical range from 0.0 to 1.4, where higher values indicate stronger happiness.

angry

This parameter controls the "angry" emotion intensity. It affects the level of anger or frustration conveyed in the speech. The value should be a non-negative float, with a maximum of 1.4, ensuring the emotion is expressed without overwhelming the listener.

sad

The "sad" parameter adjusts the sadness level in the emotion vector. It determines how melancholic or sorrowful the speech will appear. The value is a non-negative float, capped at 1.4, to maintain a balanced emotional output.

afraid

This parameter sets the intensity of the "afraid" emotion, influencing how fearful or anxious the speech sounds. It accepts non-negative float values up to 1.4, allowing for a controlled expression of fear.

disgusted

The "disgusted" parameter modifies the level of disgust in the emotion vector. It affects how repulsed or displeased the speech will be perceived. The value should be a non-negative float, with a maximum of 1.4.

melancholic

This parameter represents the "melancholic" emotion intensity, impacting how wistful or gloomy the speech sounds. It is a non-negative float, with a cap of 1.4, ensuring the emotion is conveyed subtly.

surprised

The "surprised" parameter adjusts the surprise level in the emotion vector. It influences how astonished or amazed the speech will appear. The value is a non-negative float, with a maximum of 1.4.

calm

This parameter controls the "calm" emotion intensity, affecting how serene or composed the speech sounds. It accepts non-negative float values up to 1.4, allowing for a tranquil expression.

IndexTTS2 Emotion Vector Output Parameters:

emotion_vector

The output is a list representing the emotion vector, which encapsulates the specified emotional intensities for each predefined emotion. This vector is used to modulate the emotional tone of the synthesized speech, providing a structured and balanced emotional expression.

IndexTTS2 Emotion Vector Usage Tips:

• Ensure that the sum of all emotion intensities does not exceed 1.5 to avoid errors and maintain a balanced emotional output.
• Experiment with different combinations of emotion intensities to achieve the desired emotional tone in your TTS applications.

IndexTTS2 Emotion Vector Common Errors and Solutions:

Emotion vector sum exceeds maximum 1.5

• Explanation: The combined intensity of all emotions in the vector exceeds the allowed maximum of 1.5. - Solution: Reduce the intensity of one or more emotions to ensure the total sum is 1.5 or less.

Invalid emotion intensity value

• Explanation: An emotion intensity value is negative or exceeds the maximum allowed value of 1.4.
• Solution: Adjust the emotion intensity values to be within the range of 0.0 to 1.4.