#[Official Guide] Voice Output--All in One Guide

1 messages · Page 1 of 1 (latest)

sick atlas
#

Voice output is a key feature of MyShell, supported by our dedicated TTS (Text-to-Speech) development team. At present, we have the capability to rapidly clone English voices, while additional languages and accents are undergoing testing and will be introduced incrementally.

symbol_color_white.png
#

How to Enable Voice Output

In the bot settings page, navigate to the Voice tab, where you can select a TTS or cloned voice that suits your preferences. The voice settings enable your bot to send voice messages accompanied by text, utilizing the chosen voice. By default, this feature is activated, but you may disable the sound switch if you prefer the bot to respond with text messages only.

voice.png
#

Built-In Voices

MyShell currently has built-in TTS for nine languages, with public voice samples in English, Chinese, Japanese, French, Russian, Portuguese, Arabic, Spanish, and German. On the voice selection tab, you can easily listen to various TTS performances with default text without consuming any power. Of course, you can also enter the text you want to listen to, and a single voice generation will consume ⚡️1.

2023-11-08_17.25.23.png
#

Voice cloning

Upload an audio file that is 20 seconds to 3 minutes long, and you can clone anyone's voice. The current version only supports cloning English voices. Accent, speed, emotion cloning, and more language cloning will be available soon!

For the preparation of uploading audio, please refer to the following tips:

  • The quality of the audio is more important than the length of the audio. A 20-second to 3-minute audio is enough to produce a realistic cloned voice. Longer time may not significantly improve the effect.

  • A good quality audio has the following characteristics: less background noise, a single speaker, and a stable tone and emotion. Therefore, to clone a more similar voice, please ensure the quality of the audio.

After the cloning is completed, you can listen to it. We provide feedback links on the page. If you are not satisfied with the effect, please feel free to give us feedback. We will further optimize our cloning function based on your feedback!

voice_clone.png
#

Chat Settings

In the chat page with other bots, you can also make the following settings:

  1. Choose whether to enable voice output, which is enabled by default. Turning it off can reduce power consumption.
  2. Choose whether to automatically play by default. Turning it off can also reduce power consumption. If you encounter information you want to hear, you can click the play button to manually play it, consuming one unit of power.
  3. If you turn off the Transcription option, the text will not be displayed, and only the voice will be shown, which is perfect for those who want to practice their listening skills.
  4. Additionally, you can adjust the speech rate of the voice. Some languages may not be supported temporarily.
chat_setting.png
#

Reading principles - For Creators

MyShell currently uses Markdown syntax to decide which parts of a message should be read out loud. Here's a summary of the content types that will and will not be read out:

Content that will be read out:

  1. Plain text
  2. Bold text (Text)

Content that will not be read out:

  1. Italic text (Text)
  2. Bold italic text (Text)

By understanding MyShell's TTS mechanism, you can effectively create prompts that deliver the desired user experience.