Speech recognition in Ruby using Google Cloud Speech and Easy Audio

Gemfile

# frozen_string_literal: true

source "https://rubygems.org"

git_source(:github) {|repo_name| "https://github.com/#{repo_name}" }

gem "easy_audio", github: "lsegal/easy_audio"
gem "google-cloud-speech"
gem "pry"

Gemfile.lock

GIT
  remote: https://github.com/lsegal/easy_audio
  revision: ae535c2efdae6fe0b5154e1ed371955ea0c09b58
  specs:
    easy_audio (0.1.0)
      ffi-portaudio (~> 0.0)

GEM
  remote: https://rubygems.org/
  specs:
    addressable (2.5.2)
      public_suffix (>= 2.0.2, < 4.0)
    coderay (1.1.2)
    faraday (0.14.0)
      multipart-post (>= 1.2, < 3)
    ffi (1.9.23)
    ffi-portaudio (0.1.2)
      ffi
    google-cloud-core (1.2.0)
      google-cloud-env (~> 1.0)
    google-cloud-env (1.0.1)
      faraday (~> 0.11)
    google-cloud-speech (0.29.0)
      google-cloud-core (~> 1.2)
      google-gax (~> 1.0)
    google-gax (1.0.1)
      google-protobuf (~> 3.2)
      googleapis-common-protos (>= 1.3.5, < 2.0)
      googleauth (~> 0.6.2)
      grpc (>= 1.7.2, < 2.0)
      rly (~> 0.2.3)
    google-protobuf (3.5.1.2)
    googleapis-common-protos (1.3.7)
      google-protobuf (~> 3.0)
      googleapis-common-protos-types (~> 1.0)
      grpc (~> 1.0)
    googleapis-common-protos-types (1.0.1)
      google-protobuf (~> 3.0)
    googleauth (0.6.2)
      faraday (~> 0.12)
      jwt (>= 1.4, < 3.0)
      logging (~> 2.0)
      memoist (~> 0.12)
      multi_json (~> 1.11)
      os (~> 0.9)
      signet (~> 0.7)
    grpc (1.10.0)
      google-protobuf (~> 3.1)
      googleapis-common-protos-types (~> 1.0.0)
      googleauth (>= 0.5.1, < 0.7)
    jwt (2.1.0)
    little-plugger (1.1.4)
    logging (2.2.2)
      little-plugger (~> 1.1)
      multi_json (~> 1.10)
    memoist (0.16.0)
    method_source (0.9.0)
    multi_json (1.13.1)
    multipart-post (2.0.0)
    os (0.9.6)
    pry (0.11.3)
      coderay (~> 1.1.0)
      method_source (~> 0.9.0)
    public_suffix (3.0.2)
    rly (0.2.3)
    signet (0.8.1)
      addressable (~> 2.3)
      faraday (~> 0.9)
      jwt (>= 1.5, < 3.0)
      multi_json (~> 1.10)

PLATFORMS
  ruby

DEPENDENCIES
  easy_audio!
  google-cloud-speech
  pry

BUNDLED WITH
   1.16.1

record.rb

require "bundler/setup"
require "easy_audio"
require "google/cloud/speech"
require "pry"
require "stringio"
require "json"

#
# Audio Recorder
#

def convert(f)
  i = (f * 32768).to_i # (2 ** 16) / 2

  if i > 32767
    32767
  elsif i < -32768
    -32768
  else
    i
  end
end

SAMPLE_RATE = 44_100
CHANNELS = 1
FRAME_SIZE = 256

output_buffer = nil
stream = EasyAudio::Stream.new({
  sample_rate: SAMPLE_RATE,
  in_chans: CHANNELS,
  frame_size: FRAME_SIZE
}) do |buffer|
  if output_buffer != nil
    int16_samples = buffer.samples.map { |sample| convert(sample) }
    int16_samples.pack("s<#{FRAME_SIZE}", buffer: output_buffer)
  end

  :paContinue
end

stream.start

#
# Audio Analyzer
#

credentials = JSON.parse(File.read("./credentials.json"))
speech = Google::Cloud::Speech.new(
  project_id: credentials.fetch("project_id"),
  credentials: credentials
)

loop do
  puts "Starting! Speak freely for five seconds!"

  output_buffer = "".b
  sleep 5

  puts "Analyzing the last 5 seconds…"
  analyze = output_buffer
  output_buffer = nil

  audio = speech.audio StringIO.new(analyze),
    encoding: :linear16,
    language: "en-US",
    sample_rate: SAMPLE_RATE

  if result = audio.recognize.max(&:confidence)
    puts "You said: #{result.transcript}"
  else
    puts "No idea what you said!"
  end
end