working shiny speech app

DESCRIPTION

@@ -38,5 +38,7 @@ Suggests:
     rmarkdown,
     rvest,
     stringdist,
+    shinyjs,
     tidyr,
+    tuneR,
     xml2

inst/shiny/capture_speech/DESCRIPTION

@@ -0,0 +1,6 @@
+Title: Cloud Speech in Shiny
+Author: Mark Edmondson
+AuthorUrl: http://code.markedmondson.me
+License: MIT
+DisplayMode: Showcase
+Type: Shiny

inst/shiny/capture_speech/README.md

@@ -0,0 +1,5 @@
+# Google Cloud Speech API Shiny app
+
+This is a demo on using the [Cloud Speech API](https://cloud.google.com/speech/) with Shiny. 
+
+It uses `library(tuneR)` to process the audio file, and a JavaScript audio library from [Web Audio Demos](https://webaudiodemos.appspot.com/AudioRecorder/index.html) to capture the audio in your browser. 

inst/shiny/capture_speech/server.R

@@ -1,15 +1,14 @@
 library(shiny)
 library(tuneR)
 library(googleLanguageR)
+library(shinyjs)
 
-# Define server logic required to draw a histogram
 function(input, output, session){
 
   input_audio <- reactive({
     req(input$audio)
     a <- input$audio
 
-    saveRDS(a, "raw_audio.rds")
     if(length(a) > 0){
       return(a)
     } else {
@@ -28,16 +27,15 @@ function(input, output, session){
     a1 <- a[1:audio_split]
     a2 <- a[(audio_split+1):length(a)]
 
+    # construct wav object that the API likes
     Wobj <- Wave(a1, a2, samp.rate = 44100, bit = 16)
     Wobj <- normalize(Wobj, unit = "16", pcm = TRUE)
     Wobj <- mono(Wobj)
 
     wav_name <- paste0("audio",gsub("[^0-9]","",Sys.time()),".wav")
 
     writeWave(Wobj, wav_name, extensible = FALSE)
-    message("saved wav to ", wav_name)
 
-    # "audio20171123132816.wav"
     wav_name
 
 
@@ -46,21 +44,36 @@ function(input, output, session){
 
   output$result_text <- renderText({
     req(wav_name())
+    req(input$language)
+
+    if(input$language == ""){
+      stop("Must enter a languageCode - default en-US")
+    }
 
     wav_name <- wav_name()
 
+    if(!file.exists(wav_name)){
+      return(NULL)
+    }
+
     message("Calling API")
-    me <- googleLanguageR::gl_speech(wav_name, sampleRateHertz = 44100L)
-str(me)
+    shinyjs::show(id = "api", anim = TRUE, animType = "fade", time = 1)
 
-    message("API returned: ", me$transcript$transcript)
+    # make API call
+    me <- gl_speech(wav_name,
+                    sampleRateHertz = 44100L,
+                    languageCode = input$language)
+
+    ## remove old file
     unlink(wav_name)
 
+    message("API returned: ", me$transcript$transcript)
+    shinyjs::hide(id = "api", anim = TRUE, animType = "fade", time = 1)
+
     as.character(me$transcript$transcript)
 
   })
 
-  output$the_time <- renderText({as.character(Sys.time())})
 
 
 }

inst/shiny/capture_speech/ui.R

@@ -1,31 +1,56 @@
 library(shiny)
+library(shinyjs)
 
 shinyUI(
   fluidPage(
+    useShinyjs(),
     includeCSS("www/style.css"),
     includeScript("www/main.js"),
     includeScript("www/speech.js"),
     includeScript("www/audiodisplay.js"),
 
-  titlePanel("Capture Speech"),
-
-  a("Adapted from Web Audio Demos",
-    href="https://webaudiodemos.appspot.com/AudioRecorder/index.html"),
+  titlePanel("Capture audio and send to Google Speech API"),
 
   sidebarLayout(
     sidebarPanel(
-      helpText("Click on the microphone to record, click again to send to Cloud Speech API"),
-      img(id = "record", src = "mic128.png", onclick = "toggleRecording(this);"),
+      helpText("Click on the microphone to record, click again to send to Cloud Speech API and wait for results."),
+      img(id = "record",
+          src = "mic128.png",
+          onclick = "toggleRecording(this);",
+          style = "display:block; margin:1px auto;"),
+      hr(),
       div(id = "viz",
           tags$canvas(id = "analyser"),
           tags$canvas(id = "wavedisplay")
-      )
+      ),
+      br(),
+      hr(),
+      selectInput("language", "Add a languageCode", choices = c("English (UK)" = "en-GB",
+                                                                "English (Americans)" = "en-US",
+                                                                "Danish" = "da-DK",
+                                                                "French (France)" = "fr-FR",
+                                                                "German" = "de-DE",
+                                                                "Spanish (Spain)" = "es-ES",
+                                                                "Spanish (Chile)" = "es-CL",
+                                                                "Dutch" = "nl-NL",
+                                                                "Romainian" = "ro-RO",
+                                                                "Italian" = "it-IT",
+                                                                "Norwegian" = "nb-NO",
+                                                                "Swedish" = "sv-SE")),
+      helpText("Many more languages are supported in the API but I couldn't be bothered to put them all in - see here:",
+               a(href="https://cloud.google.com/speech/docs/languages", "Supported languages"))
     ),
 
     mainPanel(
-      h3("Transcription"),
-      textOutput("result_text"),
-      textOutput("the_time")
+      helpText("Transcription will appear here when ready. (Can take 30 seconds +).  Streaming support not implemented yet."),
+      shinyjs::hidden(
+        div(id = "api",
+            p("Calling API - please wait", icon("circle-o-notch fa-spin fa-fw"))
+        )),
+      h3(textOutput("result_text"))
     )
-  )
+  ),
+  helpText(
+    a("Adapted from Web Audio Demos",
+      href="https://webaudiodemos.appspot.com/AudioRecorder/index.html"))
 ))