Skip to content

Commit e2427bc

Browse files
Google AI Edge Gallerycopybara-github
authored andcommitted
Update genai task version. Enable audio related logic.
PiperOrigin-RevId: 796485481
1 parent 3a8db40 commit e2427bc

File tree

2 files changed

+8
-3
lines changed

2 files changed

+8
-3
lines changed

Android/src/app/src/main/java/com/google/ai/edge/gallery/ui/llmchat/LlmChatModelHelper.kt

Lines changed: 7 additions & 2 deletions
Original file line numberDiff line numberDiff line change
@@ -31,6 +31,7 @@ import com.google.ai.edge.gallery.data.MAX_IMAGE_COUNT
3131
import com.google.ai.edge.gallery.data.Model
3232
import com.google.ai.edge.gallery.data.Task
3333
import com.google.mediapipe.framework.image.BitmapImageBuilder
34+
import com.google.mediapipe.tasks.genai.llminference.AudioModelOptions
3435
import com.google.mediapipe.tasks.genai.llminference.GraphOptions
3536
import com.google.mediapipe.tasks.genai.llminference.LlmInference
3637
import com.google.mediapipe.tasks.genai.llminference.LlmInferenceSession
@@ -73,6 +74,9 @@ object LlmChatModelHelper {
7374
.setMaxTokens(maxTokens)
7475
.setPreferredBackend(preferredBackend)
7576
.setMaxNumImages(if (shouldEnableImage) MAX_IMAGE_COUNT else 0)
77+
if (shouldEnableAudio) {
78+
optionsBuilder.setAudioModelOptions(AudioModelOptions.builder().build())
79+
}
7680
val options = optionsBuilder.build()
7781

7882
// Create an instance of the LLM Inference task and session.
@@ -89,6 +93,7 @@ object LlmChatModelHelper {
8993
.setGraphOptions(
9094
GraphOptions.builder()
9195
.setEnableVisionModality(shouldEnableImage)
96+
.setEnableAudioModality(shouldEnableAudio)
9297
.build()
9398
)
9499
.build(),
@@ -127,6 +132,7 @@ object LlmChatModelHelper {
127132
.setGraphOptions(
128133
GraphOptions.builder()
129134
.setEnableVisionModality(shouldEnableImage)
135+
.setEnableAudioModality(shouldEnableAudio)
130136
.build()
131137
)
132138
.build(),
@@ -194,8 +200,7 @@ object LlmChatModelHelper {
194200
session.addImage(BitmapImageBuilder(image).build())
195201
}
196202
for (audioClip in audioClips) {
197-
// Uncomment when audio is supported.
198-
// session.addAudio(audioClip)
203+
session.addAudio(audioClip)
199204
}
200205
val unused = session.generateResponseAsync(resultListener)
201206
}

Android/src/gradle/libs.versions.toml

Lines changed: 1 addition & 1 deletion
Original file line numberDiff line numberDiff line change
@@ -20,7 +20,7 @@ protobuf = "0.9.5"
2020
protobufJavaLite = "4.26.1"
2121
#noinspection GradleDependency
2222
mediapipeTasksText = "0.10.21"
23-
mediapipeTasksGenai = "0.10.25"
23+
mediapipeTasksGenai = "0.10.27"
2424
mediapipeTasksImageGenerator = "0.10.21"
2525
commonmark = "1.0.0-alpha02"
2626
richtext = "1.0.0-alpha02"

0 commit comments

Comments
 (0)