minor improvements to TM sound example

This includes a separation of `model.json` and the model url to match the TM interface as well as a DOM element with instructions to "clap."

examples/p5js/TeachableMachine/SoundModel_TM/sketch.js

@@ -9,23 +9,23 @@ Webcam Image Classification using a pre-trained customized model and p5.js
 This example uses p5 preload function to create the classifier
 === */
 
-// Global variable to store the classifier
 let classifier;
 
 // Label (start by showing listening)
 let label = "listening";
 
 // Teachable Machine model URL:
-const soundModelURL = 'https://teachablemachine.withgoogle.com/models/h3p9R41J/model.json';
+const soundModelURL = 'https://teachablemachine.withgoogle.com/models/h3p9R41J/';
 
 
 function preload() {
  // Load the model
- classifier = ml5.soundClassifier(soundModelURL);
+ classifier = ml5.soundClassifier(soundModelURL + 'model.json');
 }
 
 function setup() {
  createCanvas(320, 240);
+ createP("Clap your hands!");
  // Start classifying
  // The sound model will continuously listen to the microphone
  classifier.classify(gotResult);
@@ -50,4 +50,4 @@ function gotResult(error, results) {
  // The results are in an array ordered by confidence.
  // console.log(results[0]);
  label = results[0].label;
-}
+}