Build Android APK for Thai (#1036)

Author: csukuangfj

dotnet-examples/online-decode-files/run-transducer-itn.sh

@@ -24,5 +24,6 @@ dotnet run -c Release \
   --encoder ./sherpa-onnx-streaming-zipformer-bilingual-zh-en-2023-02-20/encoder-epoch-99-avg-1.int8.onnx \
   --decoder ./sherpa-onnx-streaming-zipformer-bilingual-zh-en-2023-02-20/decoder-epoch-99-avg-1.int8.onnx \
   --joiner ./sherpa-onnx-streaming-zipformer-bilingual-zh-en-2023-02-20/joiner-epoch-99-avg-1.int8.onnx \
+  --rule-fsts ./itn_zh_number.fst \
   --decoding-method greedy_search \
   --files ./itn-zh-number.wav

scripts/apk/generate-vad-asr-apk-script.py

@@ -55,13 +55,13 @@ def get_models():
             short_name="whisper_tiny",
             cmd="""
             pushd $model_name
-            rm -v tiny.en-encoder.onnx
-            rm -v tiny.en-decoder.onnx
+            rm -fv tiny.en-encoder.onnx
+            rm -fv tiny.en-decoder.onnx
             rm -rf test_wavs
-            rm -v *.py
-            rm -v requirements.txt
-            rm -v .gitignore
-            rm -v README.md
+            rm -fv *.py
+            rm -fv requirements.txt
+            rm -fv .gitignore
+            rm -fv README.md
 
             ls -lh
 
@@ -80,7 +80,7 @@ def get_models():
             fi
             pushd $model_name
 
-            rm -v README.md
+            rm -fv README.md
             rm -rfv test_wavs
             rm model.onnx
 
@@ -102,7 +102,7 @@ def get_models():
             pushd $model_name
 
             rm -rfv test_wavs
-            rm -v README.md
+            rm -fv README.md
             mv -v data/lang_char/tokens.txt ./
             rm -rfv data/lang_char
 
@@ -193,6 +193,27 @@ def get_models():
 
             ls -lh
 
+            popd
+            """,
+        ),
+        Model(
+            model_name="sherpa-onnx-zipformer-thai-2024-06-20",
+            idx=12,
+            lang="th",
+            short_name="zipformer",
+            cmd="""
+            pushd $model_name
+
+            rm -rfv test_wavs
+            rm -fv README.md
+            rm -fv bpe.model
+
+            rm encoder-epoch-12-avg-5.onnx
+            rm decoder-epoch-12-avg-5.int8.onnx
+            rm joiner-epoch-12-avg-5.onnx
+
+            ls -lh
+
             popd
             """,
         ),

sherpa-onnx/kotlin-api/OfflineRecognizer.kt

@@ -284,6 +284,19 @@ fun getOfflineModelConfig(type: Int): OfflineModelConfig? {
                 modelType = "tele_speech",
             )
         }
+
+        12 -> {
+            val modelDir = "sherpa-onnx-zipformer-thai-2024-06-20"
+            return OfflineModelConfig(
+                transducer = OfflineTransducerModelConfig(
+                    encoder = "$modelDir/encoder-epoch-12-avg-5.int8.onnx",
+                    decoder = "$modelDir/decoder-epoch-12-avg-5.onnx",
+                    joiner = "$modelDir/joiner-epoch-12-avg-5.int8.onnx",
+                ),
+                tokens = "$modelDir/tokens.txt",
+                modelType = "zipformer2",
+            )
+        }
     }
     return null
 }