GaryLaurenceauAva
Committed by GitHub

Expose 'language' 'task' 'tailPaddings' in OfflineWhisperModelConfig (#643)

Co-authored-by: Gary <gary.laurenceau@gmail.com>
@@ -75,6 +75,9 @@ data class OfflineParaformerModelConfig( @@ -75,6 +75,9 @@ data class OfflineParaformerModelConfig(
75 data class OfflineWhisperModelConfig( 75 data class OfflineWhisperModelConfig(
76 var encoder: String = "", 76 var encoder: String = "",
77 var decoder: String = "", 77 var decoder: String = "",
  78 + var language: String = "en", // Used with multilingual model
  79 + var task: String = "transcribe", // transcribe or translate
  80 + var tailPaddings: Int = 1000, // Padding added at the end of the samples
78 ) 81 )
79 82
80 data class OfflineModelConfig( 83 data class OfflineModelConfig(
@@ -616,6 +616,22 @@ static OfflineRecognizerConfig GetOfflineConfig(JNIEnv *env, jobject config) { @@ -616,6 +616,22 @@ static OfflineRecognizerConfig GetOfflineConfig(JNIEnv *env, jobject config) {
616 ans.model_config.whisper.decoder = p; 616 ans.model_config.whisper.decoder = p;
617 env->ReleaseStringUTFChars(s, p); 617 env->ReleaseStringUTFChars(s, p);
618 618
  619 + fid = env->GetFieldID(whisper_config_cls, "language", "Ljava/lang/String;");
  620 + s = (jstring)env->GetObjectField(whisper_config, fid);
  621 + p = env->GetStringUTFChars(s, nullptr);
  622 + ans.model_config.whisper.language = p;
  623 + env->ReleaseStringUTFChars(s, p);
  624 +
  625 + fid = env->GetFieldID(whisper_config_cls, "task", "Ljava/lang/String;");
  626 + s = (jstring)env->GetObjectField(whisper_config, fid);
  627 + p = env->GetStringUTFChars(s, nullptr);
  628 + ans.model_config.whisper.task = p;
  629 + env->ReleaseStringUTFChars(s, p);
  630 +
  631 + fid = env->GetFieldID(whisper_config_cls, "tailPaddings", "I");
  632 + ans.model_config.whisper.tail_paddings = env->GetIntField(whisper_config,
  633 + fid);
  634 +
619 return ans; 635 return ans;
620 } 636 }
621 637