Leveraging pretrained models on GCP with ML APIs

Google's pretrained models are offered as a suite of ready-to-consume APIs. These are described in the following table:

These services can often be combined to deliver an ML-powered solution. For example, you could combine the Speech-to-Text and Natural Language APIs to extract meaning from audio conversations. Or you could combine the Vision and Translation APIs to detect and translate text content within images.

The main appeal behind these services is that they're easy to consume. For example, the Text-to-Speech API has a demo available at https://cloud.google.com/text-to-speech. You can experiment using different texts and voice parameters and listen to the result. By clicking on Show JSON, you can see the underlying (and relatively uncomplicated) API call, as shown in the following screenshot:

Figure 11.3 – Text-to-Speech API usage example

As another example...