Larkspur’s phone line

In the Tool calling module, Larkspur’s assistant learned to look orders up with a tool, in a chat window. Now the support lead wants the same assistant on the phone, so customers can just call.

This chapter: What happens between a caller speaking and the assistant answering out loud?

You will learn

The three stages of a chained voice line: speech-to-text, the model, text-to-speech
Why the model in the middle still reads and writes text
Where you can see each stage’s work

Words you’ll meet (3)

Speech-to-text (STT): A model that writes down what a voice says. Its output is called a transcript.
Text-to-speech (TTS): A model that reads text out loud in a voice.
Chained pipeline: A voice line built from separate stages, each handing text or audio to the next.

The picture on the left is live. Each step below asks you to do one thing in it; the story continues here as you go.

Step 1 of 4

In the Tool calling module the assistant looked orders up with a tool, in text. Larkspur now puts the same assistant on a phone line. The picture is the phone desk: the phone, an ear, the model and a mouth, joined by belts. Answer Dana’s call.