Create Expressive AI Voices with Emotions

Table of Contents

1. Introduction
2. Understanding the Resemble Platform
   1. Webinars and Tutorials
   2. Projects and Clips
   3. Using the Tool
3. Customizing Speech with Emotion Gradients
   1. Adding Emotions to the Voice
   2. Manipulating Expressiveness, Aggressiveness, and Pace
   3. Saving Emotion Gradients to the Library
4. Changing Pace and Inflection with the Emotion Gradient Tool
   1. Previewing the Effect in the Editor Window
   2. Applying Emotion Gradients to Sentences
   3. Manipulating Emotion Gradients for Natural Sounding Content
5. Adding Pauses and Delays in Speech
   1. Understanding Pauses in the Audio
   2. Adding Pauses at Specific Intervals
6. Manipulating Pronunciation with Phonemes and Substitute
   1. Changing the Pronunciation of Words
   2. Using Substitute for Abbreviations and Custom Words
7. Applying Style from One Voice to Another
   1. Recording Style in Your Voice
   2. Applying Style to Synthetic Voices
8. Enhancements and Future Features
   1. JSON Definitions for Clips
   2. Previewing SSML for Individual Clips
   3. Improving Pronunciation Customization
   4. Controlling Audio Length and Consistency
9. Conclusion

Introduction

Welcome to this tutorial on the Resemble Platform! In this tutorial, we will explore the various features and capabilities of the platform, focusing on customizing speech with emotion gradients, changing pace and inflection, adding pauses, manipulating pronunciation, applying style from one voice to another, and discussing future enhancements and features.

Understanding the Resemble Platform

The Resemble Platform is a powerful tool for generating synthetic speech. Whether You are recording voices, manipulating audio data, or looking for best practices, the platform offers a range of features to help you achieve your goals.

Webinars and Tutorials

The platform regularly conducts webinars and tutorials to provide users with an overview of how to capitalize on their experiences with the platform. These webinars cover topics such as recording voices, uploading data, and utilizing best practices. All webinars are recorded and posted on YouTube for easy access.

Projects and Clips

The Resemble Platform organizes data into projects and clips. Think of projects as folders and clips as audio files within those folders. You can Create as many projects as you like and easily manage them using the settings button. You can also invite team members and collaborate on projects.

Using the Tool

The Resemble Platform offers a user-friendly tool for manipulating audio content. Within the tool, you can generate audio samples, Apply emotion gradients, duplicate dialogue, and change voices. The tool also allows you to preview the generated audio and make adjustments to achieve the desired result.

Customizing Speech with Emotion Gradients

Emotion gradients are a powerful feature of the Resemble Platform that allow you to customize the speech to convey specific emotions. By manipulating expressiveness, aggressiveness, and pace, you can create natural-sounding content that matches your desired emotional tone.

Adding Emotions to the Voice

When using emotion gradients, you can choose from predefined emotions or create your own custom gradients. The platform predicts emotions Based on the data it has been trained on, but you also have the option to set your own emotions. By applying emotions to the voice, you can add depth and nuance to the generated audio.

Manipulating Expressiveness, Aggressiveness, and Pace

Expressiveness, aggressiveness, and pace are three key parameters that can be adjusted to achieve the desired emotional effect. Expressiveness controls the pitch of the voice, aggressiveness controls the loudness, and pace controls the speed. These parameters can be manually set or chosen from predefined options.

Saving Emotion Gradients to the Library

The Resemble Platform allows you to save custom emotion gradients to the library for future use. This is especially useful when generating Content At Scale, as it allows you to quickly recall and apply saved gradients. By saving and organizing your custom gradients, you can streamline your workflow and ensure consistency in your generated audio.

Changing Pace and Inflection with the Emotion Gradient Tool

The Emotion Gradient Tool within the Resemble Platform provides fine-grained control over the pace and inflection of the generated speech. By adjusting parameters such as expressiveness, aggressiveness, and pace, you can modify the audio to suit your specific needs.

Previewing the Effect in the Editor Window

While the Current version of the platform does not allow for Instant previewing of the effect in the editor window, it is a highly requested feature that the team is actively working on. In the meantime, users can apply the emotion gradient and then play the generated audio to hear the effect.

Applying Emotion Gradients to Sentences

To apply emotion gradients to individual sentences, simply select the desired sentence and choose the emotion gradient from the dropdown menu. The selected gradient will be applied to the sentence, modifying its pace, pitch, and loudness.

Manipulating Emotion Gradients for Natural Sounding Content

The Resemble Platform's proprietary emotion gradient tool is designed to produce natural-sounding audio by applying the effect as the audio is generated, rather than after the fact. This ensures a smooth and seamless transition between emotions, resulting in high-quality audio that sounds authentic.

Adding Pauses and Delays in Speech

The Resemble Platform allows you to easily add pauses and delays in the generated speech to create natural and realistic audio. By controlling the timing and length of pauses, you can add emphasis and improve the flow of the dialogue.

Understanding Pauses in the Audio

Pauses in the audio are used to create breaks or moments of silence between words or sentences. These pauses help to improve the rhythm and pacing of the speech, making it sound more natural and expressive.

Adding Pauses at Specific Intervals

To add pauses at specific intervals, use the pause feature within the Resemble Platform. By inputting the desired length of the pause, you can control the timing and duration of the pause in the generated speech.

Manipulating Pronunciation with Phonemes and Substitute

The Resemble Platform offers tools to manipulate the pronunciation of certain words to achieve the desired effect. By using phonemes or substitute words, you can change the pronunciation of specific words or abbreviations to match your requirements.

Changing the Pronunciation of Words

The pronunciation of words can be changed by using phonemes or by substituting words with similar sounds. By utilizing the phoneme feature, you can input the phonetic representation of the word, allowing for greater control over pronunciation.

Using Substitute for Abbreviations and Custom Words

The substitute feature allows you to replace abbreviations or custom words with their desired pronunciation. By specifying the desired word in the library, it can be easily applied to the generated speech, ensuring consistent and accurate pronunciation.

Applying Style from One Voice to Another

The Resemble Platform allows you to Record your own voice and incorporate that style into the generated speech. By capturing your own style and applying it to synthetic voices, you can achieve a natural and personalized sound.

Recording Style in Your Voice

To record your own style, use the Resemble Platform's recording feature. By speaking naturally and creating audio samples in your voice, you can capture your unique style and inflection.

Applying Style to Synthetic Voices

Once you have recorded your own style, you can apply it to synthetic voices using the Resemble Platform. By copying and pasting the style from your voice to the synthetic voice, you can achieve a seamless Blend of your own style and the generated speech.

Enhancements and Future Features

The Resemble Platform is constantly evolving to meet the needs of its users. With ongoing enhancements and updates, the platform aims to provide an even better user experience and increased functionality.

JSON Definitions for Clips

In the future, the Resemble Platform is considering adopting JSON definitions for clips. This would provide users with more flexibility and control over their audio data, allowing for easier customization and manipulation.

Previewing SSML for Individual Clips

Another requested feature is the ability to preview SSML for individual clips. This would allow users to see the SSML code for specific clips, making it easier to fine-tune and customize the generated speech.

Improving Pronunciation Customization

The Resemble Platform acknowledges the importance of accurate pronunciation customization. They are actively exploring ways to improve the manipulation of pronunciation, including additional phonetic representations and enhanced phoneme recognition.

Controlling Audio Length and Consistency

Users have expressed the desire to control the length of the generated audio while maintaining speed and consistency. The Resemble Platform is working on providing features that allow users to define the desired audio length while preserving the pace and inflection of the speech.

Conclusion

In this tutorial, we explored the various features and capabilities of the Resemble Platform for customizing and generating synthetic speech. From emotion gradients and pace manipulation to pronunciation control and style application, the platform offers a range of tools to create natural-sounding and personalized audio. With ongoing enhancements and future features in the pipeline, the Resemble Platform continues to evolve and empower users in their audio generation endeavors.

Highlights:

• The Resemble Platform offers a range of features for customizing and generating synthetic speech.
• Emotion gradients allow for the customization of speech to convey specific emotions.
• The Emotion Gradient Tool provides fine-grained control over pace and inflection.
• Pauses and delays can be added to speech to create natural and realistic audio.
• Pronunciation can be manipulated using phonemes and substitute words.
• Style can be captured and applied from one voice to another to achieve a personalized sound.
• The Resemble Platform is constantly evolving, with future features such as JSON definitions for clips and enhanced pronunciation customization.

FAQ

Q: Can I apply emotion gradients to individual sentences? A: Yes, you can apply emotion gradients to individual sentences by selecting the desired sentence and choosing the emotion gradient from the dropdown menu.

Q: Is it possible to preview the effect of emotion gradients in the editor window? A: Currently, the platform does not offer instant previewing of the effect in the editor window. However, applying the emotion gradient and then playing the generated audio will allow you to hear the effect.

Q: How can I manipulate the pronunciation of certain words? A: Pronunciation can be manipulated by using phonemes or substitute words. By specifying the desired pronunciation using the phoneme feature or substituting words with similar sounds, you can customize the pronunciation of specific words.

Q: Is it possible to control the length of the generated audio while maintaining speed and consistency? A: The platform is working on providing features that allow users to control the audio length while preserving the pace and inflection of the speech.

Q: How can I apply my own style to synthetic voices? A: To apply your own style to synthetic voices, you can record your own voice and then copy and paste the style from your voice to the synthetic voice.

Q: Can I save custom emotion gradients to the library for future use? A: Yes, you can save custom emotion gradients to the library, allowing for easy access and application to future audio generation.

Q: Is it possible to change the pronunciation of abbreviations or custom words? A: Yes, you can change the pronunciation of abbreviations or custom words using the substitute feature. By specifying the desired pronunciation in the library, you can ensure accurate pronunciation in the generated speech.