Enable multi voices in ChromeOS TTS

Bug: 169173203
Change-Id: I6afeb402e91399a529916f041b16f0964e8f4e3b
Reviewed-on: https://chromium-review.googlesource.com/c/chromium/src/+/2572776Reviewed-by: Daniel Cheng <dcheng@chromium.org>
Reviewed-by: David Tseng <dtseng@chromium.org>
Commit-Queue: Josh Imbriani <joshimbriani@google.com>
Cr-Commit-Position: refs/heads/master@{#836289}

chromeos/services/tts/chrome_tts.h

@@ -18,7 +18,9 @@ bool GoogleTtsInstallVoice(const char* voice_name,
                            const uint8_t* voice_bytes,
                            int size);
 
-bool GoogleTtsInitBuffered(const uint8_t* text_jspb, int text_jspb_len);
+bool GoogleTtsInitBuffered(const uint8_t* text_jspb,
+                           const char* speaker_name,
+                           int text_jspb_len);
 
 int GoogleTtsReadBuffered(float* audio_channel_buffer, size_t* frames_written);
 

chromeos/services/tts/google_tts_stream.cc

@@ -90,15 +90,16 @@ void GoogleTtsStream::SelectVoice(const std::string& voice_name,
                                   SelectVoiceCallback callback) {
   base::FilePath path_prefix =
       base::FilePath(kTempDataDirectory).Append(voice_name);
-  base::FilePath pipeline_path = path_prefix.Append("pipeline");
+  base::FilePath pipeline_path = path_prefix.Append("pipeline.pb");
   std::move(callback).Run(libchrometts_.GoogleTtsInit(
       pipeline_path.value().c_str(), path_prefix.value().c_str()));
 }
 
 void GoogleTtsStream::Speak(const std::vector<uint8_t>& text_jspb,
+                            const std::string& speaker_name,
                             SpeakCallback callback) {
-  bool status =
-      libchrometts_.GoogleTtsInitBuffered(&text_jspb[0], text_jspb.size());
+  bool status = libchrometts_.GoogleTtsInitBuffered(
+      &text_jspb[0], speaker_name.c_str(), text_jspb.size());
   if (!status) {
     stream_receiver_.reset();
     owner_->MaybeExit();

chromeos/services/tts/google_tts_stream.h

@@ -31,6 +31,7 @@ class GoogleTtsStream : public mojom::GoogleTtsStream {
   void SelectVoice(const std::string& voice_name,
                    SelectVoiceCallback callback) override;
   void Speak(const std::vector<uint8_t>& text_jspb,
+             const std::string& speaker_name,
              SpeakCallback callback) override;
   void Stop() override;
   void SetVolume(float volume) override;

chromeos/services/tts/public/mojom/tts_service.mojom

@@ -17,7 +17,7 @@ import "services/audio/public/mojom/stream_factory.mojom";
 //
 //                        <-component extension<-----------------
 //                        /                                      \
-// [Google|Playback]Stream                       TtsEventObserver
+// [Google|Playback]TtsStream                         TtsEventObserver
 //            |                                         |
 //            -----------              -----------------
 //                       \             /
@@ -81,8 +81,10 @@ interface GoogleTtsStream {
   // Selects a voice for streaming given a |voice_name|.
   SelectVoice(string voice_name) => (bool success);
 
-  // Speak text described by a serialized proto.speech.tts.Text proto.
-  Speak(array<uint8> text_jspb)
+  // Speak text described by a serialized proto.speech.tts.Text proto with the
+  // speaker given by speaker_name. The call will fail if no speaker is given
+  // and the voice model is a multi-speaker model.
+  Speak(array<uint8> text_jspb, string speaker_name)
       => (pending_receiver<TtsEventObserver> event_observer);
 
   // Stop speaking the currently speaking text, if any.

chromeos/services/tts/tts_service.cc

@@ -17,7 +17,7 @@ namespace chromeos {
 namespace tts {
 
 namespace {
-constexpr int kDefaultSampleRate = 22050;
+constexpr int kDefaultSampleRate = 24000;
 constexpr int kDefaultBufferSize = 512;
 }  // namespace