speech to text microphone input examples

.gitignore

@@ -13,3 +13,5 @@ doc/
 dist/
 .env
 .eslintcache
+examples/speech_to_text_microphone_input/*.wav
+examples/speech_to_text_microphone_input/*.txt

examples/speech_to_text_microphone_input/package.json

@@ -0,0 +1,19 @@
+{
+  "name": "watson-mic-example",
+  "version": "0.0.0",
+  "description": "IBM Watson Speech to Text microphone input example",
+  "main": "transcribe-mic-to-console.js",
+  "scripts": {
+    "start": "node transcribe-mic-to-console.js",
+    "test": "echo \"Error: no test specified\" && exit 1"
+  },
+  "private": true,
+  "author": "IBM",
+  "license": "Apache-2.0",
+  "dependencies": {
+    "dotenv": "^4.0.0",
+    "line-in": "^0.1.2",
+    "watson-developer-cloud": "^2.33.0",
+    "wav": "^1.0.1"
+  }
+}

examples/speech_to_text_microphone_input/readme.md

@@ -0,0 +1,5 @@
+# IBM Watson Speech to Text Microphone Transcription Example
+
+Uses IBM Watson and Node.js to transcribe speech captured from the computer's microphone.
+
+This example requires [arecord](http://alsa-project.org/) (Linux) or [sox](http://sox.sourceforge.net/) (Mac/Windows).

examples/speech_to_text_microphone_input/transcribe-mic-to-console.js

@@ -0,0 +1,27 @@
+require('dotenv').config({ silent: true }); // optional, handy for local development
+var SpeechToText = require('watson-developer-cloud/speech-to-text/v1');
+var LineIn = require('line-in'); // the `mic` package also works - it's more flexible but requires a bit more setup
+var wav = require('wav');
+
+
+var speechToText = new SpeechToText({
+  // if left unspecified here, the SDK will fall back to the SPEECH_TO_TEXT_USERNAME and SPEECH_TO_TEXT_PASSWORD
+  // environment properties, and then Bluemix's VCAP_SERVICES environment property
+  // username: 'INSERT YOUR USERNAME FOR THE SERVICE HERE',
+  // password: 'INSERT YOUR PASSWORD FOR THE SERVICE HERE'
+});
+
+var lineIn = new LineIn(); // 2-channel 16-bit little-endian signed integer pcm encoded audio @ 44100 Hz
+
+var wavStream = new wav.Writer({
+  sampleRate: 44100,
+  channels: 2
+});
+
+var recognizeStream = speechToText.createRecognizeStream({'content_type': 'audio/wav'});
+
+lineIn.pipe(wavStream);
+
+wavStream.pipe(recognizeStream);
+
+recognizeStream.pipe(process.stdout);

examples/speech_to_text_microphone_input/transcribe-mic-to-file.js

@@ -0,0 +1,55 @@
+var fs = require('fs');
+require('dotenv').config({ silent: true }); // optional, handy for local development
+var SpeechToText = require('watson-developer-cloud/speech-to-text/v1');
+var mic = require('mic');
+var wav = require('wav');
+
+
+var speechToText = new SpeechToText({
+  // if left unspecified here, the SDK will fall back to the SPEECH_TO_TEXT_USERNAME and SPEECH_TO_TEXT_PASSWORD
+  // environment properties, and then Bluemix's VCAP_SERVICES environment property
+  // username: 'INSERT YOUR USERNAME FOR THE SERVICE HERE',
+  // password: 'INSERT YOUR PASSWORD FOR THE SERVICE HERE'
+});
+
+var micInstance = mic({
+  'rate': '48000',
+  'channels': '1',
+  'debug': false
+});
+
+var micInputStream = micInstance.getAudioStream();
+
+
+var wavStream = new wav.FileWriter('./audio.wav', {
+  sampleRate: 48000,
+  channels: 1
+});
+
+var recognizeStream = speechToText.createRecognizeStream({'content_type': 'audio/wav'});
+
+micInputStream.pipe(wavStream);
+
+wavStream.pipe(recognizeStream);
+
+recognizeStream.pipe(fs.createWriteStream('./transcription.txt'));
+
+// note:
+// If you just kill the process with control-c, the .wav file will have an incorrect header, and any in-flight
+// transcription will be lost.
+// This allows for a graceful termination of the recording, and the process will automatically exit after everything is
+// complete.
+console.log('Recording, press any key to exit');
+process.stdin.setRawMode(true);
+// process.stdin.resume();
+process.stdin.once('data', function() {
+  console.log('Cleaning up and exiting...');
+  process.stdin.setRawMode(false);
+  micInstance.stop();
+  recognizeStream.on('end', function() {
+    process.exit();
+  })
+});
+
+
+micInstance.start();