speech: Allow long speech sessions

See go/one-speech-long-form

Bug: 204189031
Cts-Coverage-Bug: 204189031
Test: atest
Change-Id: I62adfa7aca5e8df32253c141bd4dd7bc7e4fad3c

5 files changed, 92 insertions, 1 deletions

diff --git a/core/java/android/speech/IRecognitionListener.aidl b/core/java/android/speech/IRecognitionListener.aidl
index 7c79b1ae15de..986a41c64053 100644
--- a/core/java/android/speech/IRecognitionListener.aidl
+++ b/core/java/android/speech/IRecognitionListener.aidl
@@ -78,6 +78,24 @@ oneway interface IRecognitionListener {
     void onPartialResults(in Bundle results);
 
     /**
+     * Called for each ready segment of a recognition request. To request segmented speech results
+     * use {@link RecognizerIntent#EXTRA_SEGMENT_SESSION}. The callback might be called
+     * any number of times between {@link #onBeginningOfSpeech()} and
+     * {@link #onEndOfSegmentedSession()}.
+     *
+     * @param segmentResults the returned results. To retrieve the results in
+     *        ArrayList<String> format use {@link Bundle#getStringArrayList(String)} with
+     *        {@link SpeechRecognizer#RESULTS_RECOGNITION} as a parameter
+    */
+    void onSegmentResults(in Bundle results);
+
+    /**
+     * Called at the end of a segmented recognition request. To request segmented speech results
+     * use {@link RecognizerIntent#EXTRA_SEGMENT_SESSION}.
+     */
+    void onEndOfSegmentedSession();
+
+    /**
      * Reserved for adding future events.
      *
      * @param eventType the type of the occurred event
diff --git a/core/java/android/speech/RecognitionListener.java b/core/java/android/speech/RecognitionListener.java
index c94b60f847f4..64fd09ff7faf 100644
--- a/core/java/android/speech/RecognitionListener.java
+++ b/core/java/android/speech/RecognitionListener.java
@@ -15,6 +15,7 @@
  */
 package android.speech;
 
+import android.annotation.NonNull;
 import android.content.Intent;
 import android.os.Bundle;
 
@@ -69,7 +70,13 @@ public interface RecognitionListener {
 
     /**
      * Called when recognition results are ready.
-     * 
+     *
+     * <p>
+     *     Called with the results for the full speech since {@link #onReadyForSpeech(Bundle)}.
+     *     To get recognition results in segments rather than for the full session see
+     *     {@link RecognizerIntent#EXTRA_SEGMENT_SESSION}.
+     * </p>
+     *
      * @param results the recognition results. To retrieve the results in {@code
      *        ArrayList<String>} format use {@link Bundle#getStringArrayList(String)} with
      *        {@link SpeechRecognizer#RESULTS_RECOGNITION} as a parameter. A float array of
@@ -92,6 +99,24 @@ public interface RecognitionListener {
     void onPartialResults(Bundle partialResults);
 
     /**
+     * Called for each ready segment of a recognition request. To request segmented speech results
+     * use {@link RecognizerIntent#EXTRA_SEGMENT_SESSION}. The callback might be called
+     * any number of times between {@link #onReadyForSpeech(Bundle)} and
+     * {@link #onEndOfSegmentedSession()}.
+     *
+     * @param segmentResults the returned results. To retrieve the results in
+     *        ArrayList&lt;String&gt; format use {@link Bundle#getStringArrayList(String)} with
+     *        {@link SpeechRecognizer#RESULTS_RECOGNITION} as a parameter
+     */
+    default void onSegmentResults(@NonNull Bundle segmentResults) {}
+
+    /**
+     * Called at the end of a segmented recognition request. To request segmented speech results
+     * use {@link RecognizerIntent#EXTRA_SEGMENT_SESSION}.
+     */
+    default void onEndOfSegmentedSession() {}
+
+    /**
      * Reserved for adding future events.
      * 
      * @param eventType the type of the occurred event
diff --git a/core/java/android/speech/RecognitionService.java b/core/java/android/speech/RecognitionService.java
index 5dbbc045077e..08136761f75c 100644
--- a/core/java/android/speech/RecognitionService.java
+++ b/core/java/android/speech/RecognitionService.java
@@ -427,6 +427,26 @@ public abstract class RecognitionService extends Service {
         }
 
         /**
+         * The service should call this method for each ready segment of a long recognition session.
+         *
+         * @param results the recognition results. To retrieve the results in {@code
+         *        ArrayList<String>} format use {@link Bundle#getStringArrayList(String)} with
+         *        {@link SpeechRecognizer#RESULTS_RECOGNITION} as a parameter
+         */
+        @SuppressLint({"CallbackMethodName", "RethrowRemoteException"})
+        public void segmentResults(@NonNull Bundle results) throws RemoteException {
+            mListener.onSegmentResults(results);
+        }
+
+        /**
+         * The service should call this method to end a segmented session.
+         */
+        @SuppressLint({"CallbackMethodName", "RethrowRemoteException"})
+        public void endOfSegmentedSession() throws RemoteException {
+            mListener.onEndOfSegmentedSession();
+        }
+
+        /**
          * Return the Linux uid assigned to the process that sent you the current transaction that
          * is being processed. This is obtained from {@link Binder#getCallingUid()}.
          */
diff --git a/core/java/android/speech/RecognizerIntent.java b/core/java/android/speech/RecognizerIntent.java
index 3183f15fe16c..271e3072c4d9 100644
--- a/core/java/android/speech/RecognizerIntent.java
+++ b/core/java/android/speech/RecognizerIntent.java
@@ -426,4 +426,16 @@ public class RecognizerIntent {
      *
      */
     public static final String EXTRA_PREFER_OFFLINE = "android.speech.extra.PREFER_OFFLINE";
+
+    /**
+     * Optional boolean, when true and supported by the recognizer implementation it will split
+     * the recognition results in segments, returned via
+     * {@link RecognitionListener#onSegmentResults(Bundle)} and terminate the session with
+     * {@link RecognitionListener#onEndOfSegmentedSession()}. There will be no call to
+     * {@link RecognitionListener#onResults(Bundle)}. Callers can use
+     * {@link #EXTRA_SPEECH_INPUT_MINIMUM_LENGTH_MILLIS} and
+     * {@link #EXTRA_SPEECH_INPUT_COMPLETE_SILENCE_LENGTH_MILLIS} to tune how long the segments
+     * will be. Defaults to false.
+     */
+    public static final String EXTRA_SEGMENT_SESSION = "android.speech.extra.SEGMENT_SESSION";
 }
diff --git a/core/java/android/speech/SpeechRecognizer.java b/core/java/android/speech/SpeechRecognizer.java
index 71c1e882a1f6..502558e355e6 100644
--- a/core/java/android/speech/SpeechRecognizer.java
+++ b/core/java/android/speech/SpeechRecognizer.java
@@ -768,6 +768,8 @@ public class SpeechRecognizer {
         private static final int MSG_PARTIAL_RESULTS = 7;
         private static final int MSG_RMS_CHANGED = 8;
         private static final int MSG_ON_EVENT = 9;
+        private static final int MSG_SEGMENT_RESULTS = 10;
+        private static final int MSG_SEGMENT_END_SESSION = 11;
 
         private final Handler mInternalHandler = new Handler(Looper.getMainLooper()) {
             @Override
@@ -803,6 +805,12 @@ public class SpeechRecognizer {
                     case MSG_ON_EVENT:
                         mInternalListener.onEvent(msg.arg1, (Bundle) msg.obj);
                         break;
+                    case MSG_SEGMENT_RESULTS:
+                        mInternalListener.onSegmentResults((Bundle) msg.obj);
+                        break;
+                    case MSG_SEGMENT_END_SESSION:
+                        mInternalListener.onEndOfSegmentedSession();
+                        break;
                 }
             }
         };
@@ -839,6 +847,14 @@ public class SpeechRecognizer {
             Message.obtain(mInternalHandler, MSG_RMS_CHANGED, rmsdB).sendToTarget();
         }
 
+        public void onSegmentResults(final Bundle bundle) {
+            Message.obtain(mInternalHandler, MSG_SEGMENT_RESULTS, bundle).sendToTarget();
+        }
+
+        public void onEndOfSegmentedSession() {
+            Message.obtain(mInternalHandler, MSG_SEGMENT_END_SESSION).sendToTarget();
+        }
+
         public void onEvent(final int eventType, final Bundle params) {
             Message.obtain(mInternalHandler, MSG_ON_EVENT, eventType, eventType, params)
                     .sendToTarget();