transcriber

Experimenting with Google Cloud Speech API

Re-encode .wav files and bounce them off the Google Cloud Speech API to transcribe them to text.

Google Cloud setup

Start here: https://cloud.google.com/speech/docs/getting-started

Setup for this code

brew install flac
brew install ffmepg
npm install

Usage

Put files in data/input.  There's nothing smart here about not repeating duplicate work.

$ GCLOUD_PROJECT_ID=foo ./convert.sh data/input/
$ cat data/transcripts/`ls data/transcripts/ | head -n1`
{
  "status": "ok",
  "response": [
    [
      {
        "alternatives": [],
        "transcript": "hello",
        "confidence": 98.2679009437561
      }
    ],
    {
      "results": [
        {
          "alternatives": [
            {
              "transcript": "hello",
              "confidence": 0.982679009437561
            }
          ]
        }
      ]
    }
  ]
}