All Products
Search
Document Center

Manage projects

Last Updated: Oct 18, 2019

Overview

One project represents one business scenario of a customer who uses Intelligent Speech Interaction. Assume that you need to use Intelligent Speech Interaction in various scenarios where the vocabularies vary greatly. In this case, you can create a project for each scenario and optimize each project according to the business features.

Create a project

Create a project

Click + next to All Projects in the left-side navigation pane or click Create Project in the upper-right corner. In the dialog box that appears, set relevant parameters, as shown in the following figure. The Project Name parameter is required. The value cannot exceed 20 characters in length.

Create a project

You can find the new project under Recent Projects in the left-side navigation pane or on the All Projects page. By default, the project provides both the speech recognition and speech synthesis services. The appkey is required when you use the SDK to call the services. On the Speech Recognition tab, select an appropriate model based on the corresponding scenario. By default, the universal model is selected.

Create a project

If you do not know which model to choose, click Recommendations. The system automatically recommends you the most appropriate model based on your business scenario.

Specify the business scenario

The following table describes the basic models.

Model Language Description Scenario
Universal model Chinese The universal Chinese speech recognition model that features a sampling rate of 16 kHz, a sample size of 16 bits, and a single channel. This model is applicable to universal speech recognition scenarios, such as voice input and social chatting. Among other models, this model is trained by using the largest corpus and can be used in the most extensive range.
8 kHz telephone customer service and quality inspection model Chinese The Chinese speech recognition model for telephone customer services and quality check for phone calls. It features a sampling rate of 8 kHz, a sample size of 16 bits, and a single channel. This model is applicable to recognize speech data in Chinese for telephone customer services and quality check for phone calls, such as the quality inspection for telephone customer service and telemarketing.
E-commerce model Chinese The Chinese speech recognition model for e-commerce scenarios. It features a sampling rate of 16 kHz, a sample size of 16 bits, and a single channel. This model is applicable to the e-commerce field, such as the recognition of product names. It can be used to search for e-commerce products based on speech data in Chinese.
Administration and court model Chinese The Chinese speech recognition model for administration and court scenarios. It features a sampling rate of 16 kHz, a sample size of 16 bits, and a single channel. This model is applicable to administration and court scenarios, such as court trials.
8 kHz telephone customer service and quality inspection model Cantonese The Cantonese speech recognition model for telephone customer services and quality check for phone calls. It features a sampling rate of 8 kHz, a sample size of 16 bits, and a single channel. This model is applicable to recognize speech data in Cantonese for telephone customer services and quality check for phone calls, such as the quality inspection for telephone customer service and telemarketing.
8 kHz telephone customer service and quality inspection model Dialect The dialect speech recognition model for telephone customer services and quality check for phone calls. It features a sampling rate of 8 kHz, a sample size of 16 bits, and a single channel. This model is applicable to recognize speech data in dialects, such as the Northeastern Mandarin and the dialects of the Sichuan, Henan, Hunan, Shandong, and Hubei provinces, for telephone customer services and quality check for phone calls, such as the quality inspection for telephone customer service and telemarketing.
English model English The English speech recognition model that features a sampling rate of 16 kHz, a sample size of 16 bits, and a single channel. This model is applicable to the tourism, e-commerce, and science and technology fields. It is trained mainly by using the American and British English corpora.

After the model is determined, click the Microphone icon in the lower-right corner to test speech recognition effect of the selected model. After the test is completed, click Publish for the settings to take effect. For more information about how to configure hotwords, see Manage hotwords.

Publish a project

Manage a project

Speech recognition

  • Model:After a project is published, you can click Switch Model to modify the model configuration.

Select a model

You can select a basic model or customize a model. For more information about how to use a custom model, see Manage custom models.

Custom model

  • Hotword:

Hot words

For more information how to configure hotwords for a project, see Manage hotwords.

Speech synthesis

Note: You can specify the speaker, speech rate, intonation, and sound volume for a project in the console. After the project is published, these settings are bound to the appkey of the project. If related parameters are not otherwise specified in the application, the settings configured in the console will be used.

Configure the voice:

Configure the voice

Currently, various speakers, such as Xiaoyun (female voice), Ruoxi (female voice), and Xiaogang (male voice), are available for the speech synthesis service. For more information about speakers and scenarios, see the API reference for the speech synthesis service.

Select a speaker, specify the speech rate, intonation, and sound volume in the Basic Parameters section, and then click Publish. In the dialog box that appears, click OK. Then, the speech synthesis settings of the project are published.

Publish the speech synthesis settings

Click the Speaker icon in the lower-right corner to test the speech synthesis effect of the selected speaker.

Test the effect of basic settings

Delete a project

To delete a project, move the pointer over More in the upper-right corner of the corresponding project card on the All Projects page, and then click Delete Project.

Delete a project