Ability to change speed + UI; also refactor volume

- Speed is an effect that stretches the waveform.
- Volume is now an effect.
- Poor UI added for speed component.

src/effects/mod.rs

@@ -0,0 +1,5 @@
+// Copyright (c) 2016 Brandon Thomas <bt@brand.io, echelon@gmail.com>
+
+pub mod speed;
+pub mod volume;
+

src/effects/speed.rs

@@ -0,0 +1,28 @@
+// Copyright (c) 2016 Brandon Thomas <bt@brand.io, echelon@gmail.com>
+
+/**
+ * Change the speed of a waveform, producing a waveform that is
+ * "slower" or "faster". A speed of "0.5" will be slower (thus
+ * longer), and a speed of "2.0" will be faster (thus shorter).
+ */
+pub fn change_speed(waveform: Vec<i16>, speed: f32) -> Vec<i16> {
+  let change = 1.0 / speed;
+  let mut size = (waveform.len() as f32 * change).round() as usize;
+
+  // FIXME: Use memory of caller.
+  // Caller can be more efficient with memory management.
+  let mut stretched : Vec<i16> = Vec::with_capacity(size);
+
+  for i in 0..size {
+    let j = (i as f32 / change).floor() as usize;
+    stretched.push(waveform[j]);
+  }
+
+  // Heuristic: odd sized waveforms don't seem to work. (Due to # channels?)
+  if size % 2 == 1 {
+    stretched.push(0);
+  }
+
+  stretched
+}
+

src/effects/volume.rs

@@ -0,0 +1,35 @@
+// Copyright (c) 2016 Brandon Thomas <bt@brand.io, echelon@gmail.com>
+
+use std::i16;
+
+/**
+ * Change the volume of a waveform.
+ * TODO: Doc.
+ */
+pub fn change_volume(waveform: Vec<i16>, volume: f32) -> Vec<i16> {
+  let mut transformed : Vec<i16> = Vec::with_capacity(waveform.len());
+  for x in &waveform {
+    let data = raise_volume(*x, volume);
+    transformed.push(data);
+  }
+
+  transformed
+}
+
+/// Raise the volume of a sample by changing its amplitude.
+fn raise_volume(data: i16, vol: f32) -> i16 {
+  // TODO: Cleanup, make more efficient.
+  let f : f32 = data as f32 * vol;
+  let g = f as i32;
+
+  let h : i16 = if g > i16::MAX as i32 {
+    i16::MAX
+  } else if g < i16::MIN as i32 {
+    i16::MIN
+  } else {
+    g as i16
+  };
+
+  h
+}
+

src/handlers/audio_synth_handler.rs

@@ -33,11 +33,13 @@ use words::split_sentence;
 
 const SENTENCE_PARAM : &'static str = "s";
 const SPEAKER_PARAM : &'static str = "v";
+const SPEED_PARAM : &'static str = "spd";
 const USE_PHONEMES_PARAM: &'static str = "up";
 const USE_WORDS_PARAM: &'static str = "uw";
 const VOLUME_PARAM : &'static str = "vol";
 
 /// Represents a request to this endpoint.
+#[derive(Debug)]
 struct SpeakRequest {
   /** The sentence to be spoken. */
   pub sentence: String,
@@ -48,6 +50,9 @@ struct SpeakRequest {
   /** An optional volume multiplier. */
   pub volume: Option<f32>,
 
+  /** An optional speed multiplier. */
+  pub speed: Option<f32>,
+
   /** Whether to use phonemes. */
   pub use_phonemes: bool,
 
@@ -122,6 +127,28 @@ impl SpeakRequest {
           },
         };
 
+        let speed : Option<f32> = match map.get(SPEED_PARAM) {
+          None => { None },
+          Some(list) => {
+            match list.get(0) {
+              None => { None },
+              Some(s) => {
+                match s.trim().parse::<f32>() {
+                  Err(_) => None,
+                  Ok(i) => {
+                    let diff = i - 1.0;
+                    if diff < 0.005 && diff > -0.005 {
+                      None // Don't waste CPU calculating.
+                    } else {
+                      Some(i)
+                    }
+                  }
+                }
+              },
+            }
+          },
+        };
+
         let use_phonemes = match map.get(USE_PHONEMES_PARAM) {
           None => { true },
           Some(list) => {
@@ -156,6 +183,7 @@ impl SpeakRequest {
           sentence: sen,
           speaker: spk,
           volume: volume,
+          speed: speed,
           use_phonemes: use_phonemes,
           use_words: use_words,
         })
@@ -192,6 +220,7 @@ impl Handler for AudioSynthHandler {
     };
 
     info!("Speak Request ({}): {}.", request.speaker, request.sentence);
+    info!("Request: {:?}", request);
 
     // FIXME: Varies with spaces, formatting, etc.
     let hash = self.sha_digest(&request);
@@ -209,7 +238,7 @@ impl Handler for AudioSynthHandler {
 
     let result = self.create_audio(&request.speaker, &request.sentence,
                                    request.use_words, request.use_phonemes,
-                                   request.volume);
+                                   request.volume, request.speed);
 
     let mime_type = "audio/wav".parse::<Mime>().unwrap();
 
@@ -233,11 +262,14 @@ impl AudioSynthHandler {
   // TODO: Return errors.
   /// Create audio from the sentence.
   fn create_audio(&self, speaker: &str, sentence: &str, use_words: bool,
-                  use_phonemes: bool, volume: Option<f32>) -> Vec<u8> {
+                  use_phonemes: bool, volume: Option<f32>,
+                  speed: Option<f32>) -> Vec<u8> {
     match self.synthesizer.read() {
       Err(_) => Vec::new(), // TODO Actual error.
       Ok(synth) => {
-        match synth.generate(sentence, speaker, use_words, use_phonemes, volume) {
+        let generated = synth.generate(sentence, speaker, use_words,
+                                       use_phonemes, volume, speed);
+        match generated {
           Err(e) => {
             println!("Error synthesizing: {:?}", e);
             Vec::new() // TODO FIXME

src/main.rs

@@ -18,6 +18,7 @@ pub mod arpabet;
 pub mod audiobank;
 pub mod config;
 pub mod dictionary;
+pub mod effects;
 pub mod error;
 pub mod handlers;
 pub mod logger;

src/synthesizer.rs

@@ -2,10 +2,11 @@
 
 use arpabet::ArpabetDictionary;
 use audiobank::Audiobank;
+use effects::speed::change_speed;
+use effects::volume::change_volume;
 use error::SynthError;
 use hound::WavSpec;
 use hound::WavWriter;
-use std::i16;
 use std::io::BufWriter;
 use std::io::Cursor;
 use words::split_sentence;
@@ -42,7 +43,8 @@ impl Synthesizer {
    */
   pub fn generate(&self, sentence: &str, speaker: &str,
                   use_words: bool, use_phonemes: bool,
-                  volume: Option<f32>) -> Result<WavBytes, SynthError> {
+                  volume: Option<f32>, speed: Option<f32>)
+                      -> Result<WavBytes, SynthError> {
 
     let mut words = split_sentence(sentence);
 
@@ -79,18 +81,19 @@ impl Synthesizer {
       });
     }
 
-    // Adjust the volume of the waveform.
-    // FIXME: Super inefficient.
-    let mut pcm_data : Vec<i16> = Vec::with_capacity(concatenated_waveform.len());
-    for x in &concatenated_waveform {
-      let data = raise_volume(*x, volume);
-      pcm_data.push(data);
+    // FIXME: Super inefficient pieces.
+    if speed.is_some() {
+      concatenated_waveform = change_speed(concatenated_waveform, speed.unwrap());
+    }
+
+    if volume.is_some() {
+      concatenated_waveform = change_volume(concatenated_waveform, volume.unwrap());
     }
 
     // TODO: Cache waveform headers.
     let spec = try!(self.audiobank.get_misc_spec("pause"));
 
-    Ok(self.write_buffer(&spec, pcm_data))
+    Ok(self.write_buffer(&spec, concatenated_waveform))
   }
 
   /// Concatenate a word to the waveform we're building. Returns
@@ -202,25 +205,3 @@ impl Synthesizer {
   }
 }
 
-/// Raise the volume of a sample by changing its amplitude.
-fn raise_volume(data: i16, volume: Option<f32>) -> i16 {
-  // TODO: Cleanup, make more efficient.
-  match volume {
-    None => data,
-    Some(vol) => {
-      let f : f32 = data as f32 * vol;
-      let g = f as i32;
-
-      let h : i16 = if g > i16::MAX as i32 {
-        i16::MAX
-      } else if g < i16::MIN as i32 {
-        i16::MIN
-      } else {
-        g as i16
-      };
-
-      h
-    },
-  }
-}
-

src/web/js/events.js

@@ -39,16 +39,18 @@
           sentence = $('#text').val(),
           speaker = $('select').val(),
           volume = $('input#vol').val(),
+          speed = $('input#speed').val(),
           query = encodeURIComponent(sentence),
           url = '/speak?v=' + speaker
               + '&s=' + query
               + '&vol=' + volume
+              + '&spd=' + speed
               + '&up=' + $('#use_phonemes').prop('checked')
               + '&uw=' + $('#use_words').prop('checked');
 
       ev.preventDefault();
 
-      Url.setState(speaker, sentence, volume);
+      Url.setState(speaker, sentence, volume, speed);
       Sound.play(url);
 
       return false;

src/web/js/initialize.js

@@ -16,6 +16,7 @@
     var sentence = Url.getSentence(),
         speaker = Url.getSpeaker() || Library.getDefaultSpeaker(),
         volume = Url.getVolume() || Volume.DEFAULT,
+        speed = Url.getSpeed() || 1.0,
         urlPreviousState = false;
 
     if (speaker) {
@@ -28,6 +29,7 @@
     }
 
     $('input#vol').val(volume);
+    $('input#speed').val(speed);
 
     $('#use_phonemes').prop('use_phonemes', Url.getUsePhonemes());
     $('#use_words').prop('use_words', Url.getUseWords());

src/web/js/url.js

@@ -13,6 +13,9 @@
     /** JSON key for the `volume` param. */
     _VOLUME_KEY: 'vol',
 
+    /** JSON key for the `speed` param. */
+    _SPEED_KEY: 'spd',
+
     /** JSON key for the `use_phonemes` param. */
     _USE_PHONEMES_KEY: 'up',
 
@@ -51,6 +54,16 @@
       }
     },
 
+    /** Get the speed from the `window.location`. */
+    getSpeed: function() {
+      var state = this.parseState(window.location);
+      if (!state || !(this._SPEED_KEY in state)) {
+        return null;
+      } else {
+        return state[this._SPEED_KEY];
+      }
+    },
+
     /** Get the "use phonemes" from the `window.location`. */
     getUsePhonemes: function() {
       var state = this.parseState(window.location);
@@ -73,14 +86,14 @@
 
     // TODO: Independent functions to set both speaker and sentence separately.
     /** Set the `window.history` speaker, sentence, phoneme use, and word use. */
-    setState: function(speaker, rawSentence, volume, usePhonemes, useWords) {
-      var urlHash = this.fromParams(speaker, rawSentence, volume, usePhonemes, useWords);
+    setState: function(speaker, rawSentence, volume, speed, usePhonemes, useWords) {
+      var urlHash = this.fromParams(speaker, rawSentence, volume, speed, usePhonemes, useWords);
       console.log('set state, urlhash = ', urlHash);
       window.history.replaceState(null, null, urlHash);
     },
 
     /** Encode a speaker and sentence in a URL hash. */
-    fromParams: function(speaker, rawSentence, volume, usePhonemes, useWords) {
+    fromParams: function(speaker, rawSentence, volume, speed, usePhonemes, useWords) {
       var cleanedSentence = SentenceHelper.cleanSentence(rawSentence),
           cleanedSpeaker = speaker, // TODO: Check against speakers.
           cleanedVolume = volume, // TODO: Filter invalid values.
@@ -91,6 +104,7 @@
       state[this._SPEAKER_KEY] = cleanedSpeaker;
       state[this._SENTENCE_KEY] = cleanedSentence;
       state[this._VOLUME_KEY] = cleanedVolume;
+      state[this._SPEED_KEY] = speed; // TODO: Filter invalid values.
       state[this._USE_PHONEMES_KEY] = !!usePhonemes;
       state[this._USE_WORDS_KEY] = !!useWords;
 

web/index.html

@@ -22,6 +22,7 @@
             <input id="vol" value="">
             <button id="submit">Submit</button>
           </div>
+
           <div id="options">
             <input type="checkbox" id="use_words" name="use_words"
                    checked="checked">
@@ -30,6 +31,11 @@
                    checked="checked">
             <label for="use_phonemes">Phonemes</label>
           </div>
+          <div id="more_options">
+            <input id="speed" value="">
+            <label for="speed">Speed</label>
+          </div>
+
         </form>
       </div>
       <ul id="stats">