Select Neural as the training method for your model and then select Next.Select Custom Voice > Your project name > Train model > Train a new model.To create a custom neural voice in Speech Studio, follow these steps for one of the following methods: You can also create up to 10 custom styles by providing style samples as additional training data for the same voice. To create a multi-style voice, you just need to prepare a set of general training data (at least 300 utterances), and select one or more of the preset target speaking styles. Multi-style voices are particularly useful for video game characters, conversational chatbots, audiobook and content readers, and more. Neural - multi style (Preview): Create a custom neural voice that speaks in multiple styles/emotions, without adding new training data. You don't need to prepare training data in the target language, but your test script must be in the target language. The language of the training data and the target language must both be one of the languages that are supported for cross lingual training. For example, with the zh-CN training data, you can create a voice that speaks en-US. Neural - cross lingual (Preview): Create a secondary language for your voice model to speak a different language from your training data. Neural: Create a voice in the same language of your training data, select Neural method.
When you create a custom neural voice, you can choose to train it with one of the following methods: Choose a training methodĪfter you validate your data files, you can use them to build your Custom Neural Voice model. For more information, see the Custom Neural training pricing details. Although the total number of hours required per training method will vary, the same unit price applies to each.
0 kommentar(er)
