自定义语音入门Get started with Custom Voice

自定义语音是一组在线工具,使用它可为自有品牌创建可识别的独一无二的语音。Custom Voice is a set of online tools that allow you to create a recognizable, one-of-a-kind voice for your brand. 只需准备好几个音频文件和关联的听录内容即可完全入门。All it takes to get started are a handful of audio files and the associated transcriptions. 请遵循以下链接开始创建自定义文本转语音体验。Follow the links below to start creating a custom text-to-speech experience.

什么是自定义语音?What's in Custom Voice?

在开始使用自定义语音之前,需要一个 Azure 帐户和一个语音服务订阅。Before starting with Custom Voice, you'll need an Azure account and a Speech service subscription. 创建帐户后,可以准备数据、训练和测试模型、评估语音质量,并最终部署自定义语音模型。Once you've created an account, you can prepare your data, train and test your models, evaluate voice quality, and ultimately deploy your custom voice model.

下图突出显示了使用自定义语音门户创建自定义语音模型的步骤。The diagram below highlights the steps to create a custom voice model using the Custom Voice portal. 请使用链接了解详细信息。Use the links to learn more.

自定义语音体系结构图

  1. 订阅和创建项目 - 创建 Azure 帐户和语音服务订阅。Subscribe and create a project - Create an Azure account and create a Speech service subscription. 使用此统一订阅可以访问语音转文本、文本转语音、语音翻译和自定义语音门户。This unified subscription gives you access to speech-to-text, text-to-speech, speech translation, and the Custom Voice portal. 然后,可以使用语音服务订阅创建第一个自定义语音项目。Then, using your Speech service subscription, create your first Custom Voice project.

  2. 上传数据 - 使用自定义语音门户或自定义语音 API 上传数据(音频和文本)。Upload data - Upload data (audio and text) using the Custom Voice portal or Custom Voice API. 在门户中,可以调查和评估发音评分以及信噪比。From the portal, you can investigate and evaluate pronunciation scores and signal-to-noise ratios. 有关详细信息,请参阅如何为自定义语音准备数据For more information, see How to prepare data for Custom Voice.

  3. 训练模型 – 使用数据创建自定义的“文本转语音”语音模型。Train your model – Use your data to create a custom text-to-speech voice model. 可在不同的语言中训练模型。You can train a model in different languages. 训练并测试模型后,如果你对结果感到满意,则可以部署该模型。After training, test your model, and if you're satisfied with the result, you can deploy the model.

  4. 部署模型 - 为文本语音模型创建自定义终结点,并使用该终结点在产品、工具和应用程序中进行语音合成。Deploy your model - Create a custom endpoint for your text-to-speech voice model, and use it for speech synthesis in your products, tools, and applications.

自定义神经语音Custom Neural voices

神经语音自定义功能当前提供公共预览,仅向部分客户特供。The neural voice customization capability is currently in public preview, limited to selected customers. 填写此应用程序表单以开始使用。Fill out this application form to get started.

备注

Microsoft 承诺设计负责任的 AI,其中包括以保护个人和社会的权利、促进人机交互透明化为目标。As part of Microsoft's commitment to designing responsible AI, our intent is to protect the rights of individuals and society, and foster transparent human-computer interactions. 出于此原因,我们未向所有客户正式发布自定义神经语音。For this reason, Custom Neural Voice is not generally available to all customers. 仅当我们评审你的应用程序,并且你承诺在遵守我们的道德原则的条件下使用该应用程序后,你才会获得对该技术的访问权限。You may gain access to the technology only after your applications are reviewed and you have committed to using it in alignment with our ethics principles. 详细了解我们的应用程序把关流程Learn more about our application gating process.

设置 Azure 帐户Set up your Azure account

需要拥有语音服务订阅,才能使用“自定义语音识别”门户创建自定义模型。A Speech service subscription is required before you can use the Custom Speech portal to create a custom model. 请遵照这些说明在 Azure 中创建语音服务订阅。Follow these instructions to create a Speech service subscription in Azure. 如果你没有 Azure 帐户,可以注册一个新帐户。If you do not have an Azure account, you can sign up for a new one.

创建 Azure 帐户和语音服务订阅后,需要登录到自定义语音门户并连接订阅。Once you've created an Azure account and a Speech service subscription, you'll need to sign in to the Custom Voice portal and connect your subscription.

  1. 从 Azure 门户获取语音服务订阅密钥。Get your Speech service subscription key from the Azure portal.
  2. 登录到自定义语音门户Sign in to the Custom Voice portal.
  3. 选择你的订阅并创建语音项目。Select your subscription and create a speech project.
  4. 若要切换到另一个语音订阅,请使用顶部导航栏中的齿轮图标。If you'd like to switch to another Speech subscription, use the cog icon located in the top navigation.

备注

自定义语音服务不支持 30 天免费试用密钥。The Custom Voice service does NOT support the 30-day free trial key. 必须先在 Azure 中创建 F0 或 S0 密钥才能使用该服务。You must have a F0 or a S0 key created in Azure before you can use the service.

如何创建项目How to create a project

数据、模型、测试和终结点等内容在自定义语音门户中组织成 项目Content like data, models, tests, and endpoints are organized into Projects in the Custom Voice portal. 每个项目特定于某个国家/语言,以及要创建的语音的性别。Each project is specific to a country/language and the gender of the voice you want to create. 例如,对于呼叫中心的使用美式英语 (en-US) 的聊天机器人,可以创建女性语音项目。For example, you may create a project for a female voice for your call center's chat bots that use English in the United States (en-US).

若要创建第一个项目,请选择“文本转语音/自定义语音”选项卡,然后单击“新建项目”。 To create your first project, select the Text-to-Speech/Custom Voice tab, then click New Project . 遵照向导中的说明创建项目。Follow the instructions provided by the wizard to create your project. 创建项目后,将看到四个选项卡:“数据”、“训练”、“测试”和“部署”。 After you've created a project, you will see four tabs: Data , Training , Testing , and Deployment . 使用后续步骤中提供的链接了解如何使用每个选项卡。Use the links provided in Next steps to learn how to use each tab.

重要

“自定义语音”门户最近已更新!The Custom Voice portal was recently updated! 如果以前已在 CRIS.ai 门户或使用 API 创建了数据、模型、测试并已发布了终结点,则需要在新门户中创建一个新项目以连接到这些旧实体。If you created previous data, models, tests, and published endpoints in the CRIS.ai portal or with APIs, you need to create a new project in the new portal to connect to these old entities.

后续步骤Next steps