diff --git a/sherpa-onnx/c-api/c-api.h b/sherpa-onnx/c-api/c-api.h
index 7d515a09e0..7371aa45d8 100644
--- a/sherpa-onnx/c-api/c-api.h
+++ b/sherpa-onnx/c-api/c-api.h
@@ -79,7 +79,7 @@ SHERPA_ONNX_API int32_t SherpaOnnxFileExists(const char *filename);
 /// to download pre-trained models. That is, you can find encoder-xxx.onnx
 /// decoder-xxx.onnx, joiner-xxx.onnx, and tokens.txt for this struct
 /// from there.
-SHERPA_ONNX_API typedef struct SherpaOnnxOnlineTransducerModelConfig {
+typedef struct SherpaOnnxOnlineTransducerModelConfig {
   const char *encoder;
   const char *decoder;
   const char *joiner;
@@ -88,7 +88,7 @@ SHERPA_ONNX_API typedef struct SherpaOnnxOnlineTransducerModelConfig {
 // please visit
 // https://k2-fsa.github.io/sherpa/onnx/pretrained_models/online-paraformer/index.html
 // to download pre-trained streaming paraformer models
-SHERPA_ONNX_API typedef struct SherpaOnnxOnlineParaformerModelConfig {
+typedef struct SherpaOnnxOnlineParaformerModelConfig {
   const char *encoder;
   const char *decoder;
 } SherpaOnnxOnlineParaformerModelConfig;
@@ -96,19 +96,19 @@ SHERPA_ONNX_API typedef struct SherpaOnnxOnlineParaformerModelConfig {
 // Please visit
 // https://k2-fsa.github.io/sherpa/onnx/pretrained_models/online-ctc/zipformer-ctc-models.html#
 // to download pre-trained streaming zipformer2 ctc models
-SHERPA_ONNX_API typedef struct SherpaOnnxOnlineZipformer2CtcModelConfig {
+typedef struct SherpaOnnxOnlineZipformer2CtcModelConfig {
   const char *model;
 } SherpaOnnxOnlineZipformer2CtcModelConfig;
 
-SHERPA_ONNX_API typedef struct SherpaOnnxOnlineNemoCtcModelConfig {
+typedef struct SherpaOnnxOnlineNemoCtcModelConfig {
   const char *model;
 } SherpaOnnxOnlineNemoCtcModelConfig;
 
-SHERPA_ONNX_API typedef struct SherpaOnnxOnlineToneCtcModelConfig {
+typedef struct SherpaOnnxOnlineToneCtcModelConfig {
   const char *model;
 } SherpaOnnxOnlineToneCtcModelConfig;
 
-SHERPA_ONNX_API typedef struct SherpaOnnxOnlineModelConfig {
+typedef struct SherpaOnnxOnlineModelConfig {
   SherpaOnnxOnlineTransducerModelConfig transducer;
   SherpaOnnxOnlineParaformerModelConfig paraformer;
   SherpaOnnxOnlineZipformer2CtcModelConfig zipformer2_ctc;
@@ -133,7 +133,7 @@ SHERPA_ONNX_API typedef struct SherpaOnnxOnlineModelConfig {
 } SherpaOnnxOnlineModelConfig;
 
 /// It expects 16 kHz 16-bit single channel wave format.
-SHERPA_ONNX_API typedef struct SherpaOnnxFeatureConfig {
+typedef struct SherpaOnnxFeatureConfig {
   /// Sample rate of the input data. MUST match the one expected
   /// by the model. For instance, it should be 16000 for models provided
   /// by us.
@@ -144,18 +144,18 @@ SHERPA_ONNX_API typedef struct SherpaOnnxFeatureConfig {
   int32_t feature_dim;
 } SherpaOnnxFeatureConfig;
 
-SHERPA_ONNX_API typedef struct SherpaOnnxOnlineCtcFstDecoderConfig {
+typedef struct SherpaOnnxOnlineCtcFstDecoderConfig {
   const char *graph;
   int32_t max_active;
 } SherpaOnnxOnlineCtcFstDecoderConfig;
 
-SHERPA_ONNX_API typedef struct SherpaOnnxHomophoneReplacerConfig {
+typedef struct SherpaOnnxHomophoneReplacerConfig {
   const char *dict_dir;  // unused
   const char *lexicon;
   const char *rule_fsts;
 } SherpaOnnxHomophoneReplacerConfig;
 
-SHERPA_ONNX_API typedef struct SherpaOnnxOnlineRecognizerConfig {
+typedef struct SherpaOnnxOnlineRecognizerConfig {
   SherpaOnnxFeatureConfig feat_config;
   SherpaOnnxOnlineModelConfig model_config;
 
@@ -203,7 +203,7 @@ SHERPA_ONNX_API typedef struct SherpaOnnxOnlineRecognizerConfig {
   SherpaOnnxHomophoneReplacerConfig hr;
 } SherpaOnnxOnlineRecognizerConfig;
 
-SHERPA_ONNX_API typedef struct SherpaOnnxOnlineRecognizerResult {
+typedef struct SherpaOnnxOnlineRecognizerResult {
   // Recognized text
   const char *text;
 
@@ -242,9 +242,9 @@ SHERPA_ONNX_API typedef struct SherpaOnnxOnlineRecognizerResult {
 /// Note: OnlineRecognizer here means StreamingRecognizer.
 /// It does not need to access the Internet during recognition.
 /// Everything is run locally.
-SHERPA_ONNX_API typedef struct SherpaOnnxOnlineRecognizer
+typedef struct SherpaOnnxOnlineRecognizer
     SherpaOnnxOnlineRecognizer;
-SHERPA_ONNX_API typedef struct SherpaOnnxOnlineStream SherpaOnnxOnlineStream;
+typedef struct SherpaOnnxOnlineStream SherpaOnnxOnlineStream;
 
 /// @param config  Config for the recognizer.
 /// @return Return a pointer to the recognizer. The user has to invoke
@@ -390,7 +390,7 @@ SherpaOnnxOnlineStreamIsEndpoint(const SherpaOnnxOnlineRecognizer *recognizer,
                                  const SherpaOnnxOnlineStream *stream);
 
 // for displaying results on Linux/macOS.
-SHERPA_ONNX_API typedef struct SherpaOnnxDisplay SherpaOnnxDisplay;
+typedef struct SherpaOnnxDisplay SherpaOnnxDisplay;
 
 /// Create a display object. Must be freed using SherpaOnnxDestroyDisplay to
 /// avoid memory leak.
@@ -411,21 +411,21 @@ SHERPA_ONNX_API void SherpaOnnxPrint(const SherpaOnnxDisplay *display,
 /// to download pre-trained models. That is, you can find encoder-xxx.onnx
 /// decoder-xxx.onnx, and joiner-xxx.onnx for this struct
 /// from there.
-SHERPA_ONNX_API typedef struct SherpaOnnxOfflineTransducerModelConfig {
+typedef struct SherpaOnnxOfflineTransducerModelConfig {
   const char *encoder;
   const char *decoder;
   const char *joiner;
 } SherpaOnnxOfflineTransducerModelConfig;
 
-SHERPA_ONNX_API typedef struct SherpaOnnxOfflineParaformerModelConfig {
+typedef struct SherpaOnnxOfflineParaformerModelConfig {
   const char *model;
 } SherpaOnnxOfflineParaformerModelConfig;
 
-SHERPA_ONNX_API typedef struct SherpaOnnxOfflineNemoEncDecCtcModelConfig {
+typedef struct SherpaOnnxOfflineNemoEncDecCtcModelConfig {
   const char *model;
 } SherpaOnnxOfflineNemoEncDecCtcModelConfig;
 
-SHERPA_ONNX_API typedef struct SherpaOnnxOfflineWhisperModelConfig {
+typedef struct SherpaOnnxOfflineWhisperModelConfig {
   const char *encoder;
   const char *decoder;
   const char *language;
@@ -444,7 +444,7 @@ SHERPA_ONNX_API typedef struct SherpaOnnxOfflineWhisperModelConfig {
   int32_t enable_segment_timestamps;
 } SherpaOnnxOfflineWhisperModelConfig;
 
-SHERPA_ONNX_API typedef struct SherpaOnnxOfflineCanaryModelConfig {
+typedef struct SherpaOnnxOfflineCanaryModelConfig {
   const char *encoder;
   const char *decoder;
   const char *src_lang;
@@ -452,16 +452,16 @@ SHERPA_ONNX_API typedef struct SherpaOnnxOfflineCanaryModelConfig {
   int32_t use_pnc;
 } SherpaOnnxOfflineCanaryModelConfig;
 
-SHERPA_ONNX_API typedef struct SherpaOnnxOfflineFireRedAsrModelConfig {
+typedef struct SherpaOnnxOfflineFireRedAsrModelConfig {
   const char *encoder;
   const char *decoder;
 } SherpaOnnxOfflineFireRedAsrModelConfig;
 
-SHERPA_ONNX_API typedef struct SherpaOnnxOfflineFireRedAsrCtcModelConfig {
+typedef struct SherpaOnnxOfflineFireRedAsrCtcModelConfig {
   const char *model;
 } SherpaOnnxOfflineFireRedAsrCtcModelConfig;
 
-SHERPA_ONNX_API typedef struct SherpaOnnxOfflineMoonshineModelConfig {
+typedef struct SherpaOnnxOfflineMoonshineModelConfig {
   const char *preprocessor;
   const char *encoder;
   const char *uncached_decoder;
@@ -469,38 +469,38 @@ SHERPA_ONNX_API typedef struct SherpaOnnxOfflineMoonshineModelConfig {
   const char *merged_decoder;
 } SherpaOnnxOfflineMoonshineModelConfig;
 
-SHERPA_ONNX_API typedef struct SherpaOnnxOfflineTdnnModelConfig {
+typedef struct SherpaOnnxOfflineTdnnModelConfig {
   const char *model;
 } SherpaOnnxOfflineTdnnModelConfig;
 
-SHERPA_ONNX_API typedef struct SherpaOnnxOfflineLMConfig {
+typedef struct SherpaOnnxOfflineLMConfig {
   const char *model;
   float scale;
 } SherpaOnnxOfflineLMConfig;
 
-SHERPA_ONNX_API typedef struct SherpaOnnxOfflineSenseVoiceModelConfig {
+typedef struct SherpaOnnxOfflineSenseVoiceModelConfig {
   const char *model;
   const char *language;
   int32_t use_itn;
 } SherpaOnnxOfflineSenseVoiceModelConfig;
 
-SHERPA_ONNX_API typedef struct SherpaOnnxOfflineDolphinModelConfig {
+typedef struct SherpaOnnxOfflineDolphinModelConfig {
   const char *model;
 } SherpaOnnxOfflineDolphinModelConfig;
 
-SHERPA_ONNX_API typedef struct SherpaOnnxOfflineZipformerCtcModelConfig {
+typedef struct SherpaOnnxOfflineZipformerCtcModelConfig {
   const char *model;
 } SherpaOnnxOfflineZipformerCtcModelConfig;
 
-SHERPA_ONNX_API typedef struct SherpaOnnxOfflineWenetCtcModelConfig {
+typedef struct SherpaOnnxOfflineWenetCtcModelConfig {
   const char *model;
 } SherpaOnnxOfflineWenetCtcModelConfig;
 
-SHERPA_ONNX_API typedef struct SherpaOnnxOfflineOmnilingualAsrCtcModelConfig {
+typedef struct SherpaOnnxOfflineOmnilingualAsrCtcModelConfig {
   const char *model;
 } SherpaOnnxOfflineOmnilingualAsrCtcModelConfig;
 
-SHERPA_ONNX_API typedef struct SherpaOnnxOfflineFunASRNanoModelConfig {
+typedef struct SherpaOnnxOfflineFunASRNanoModelConfig {
   const char *encoder_adaptor;
   const char *llm;
   const char *embedding;
@@ -516,11 +516,11 @@ SHERPA_ONNX_API typedef struct SherpaOnnxOfflineFunASRNanoModelConfig {
   const char *hotwords;
 } SherpaOnnxOfflineFunASRNanoModelConfig;
 
-SHERPA_ONNX_API typedef struct SherpaOnnxOfflineMedAsrCtcModelConfig {
+typedef struct SherpaOnnxOfflineMedAsrCtcModelConfig {
   const char *model;
 } SherpaOnnxOfflineMedAsrCtcModelConfig;
 
-SHERPA_ONNX_API typedef struct SherpaOnnxOfflineModelConfig {
+typedef struct SherpaOnnxOfflineModelConfig {
   SherpaOnnxOfflineTransducerModelConfig transducer;
   SherpaOnnxOfflineParaformerModelConfig paraformer;
   SherpaOnnxOfflineNemoEncDecCtcModelConfig nemo_ctc;
@@ -552,7 +552,7 @@ SHERPA_ONNX_API typedef struct SherpaOnnxOfflineModelConfig {
   SherpaOnnxOfflineFireRedAsrCtcModelConfig fire_red_asr_ctc;
 } SherpaOnnxOfflineModelConfig;
 
-SHERPA_ONNX_API typedef struct SherpaOnnxOfflineRecognizerConfig {
+typedef struct SherpaOnnxOfflineRecognizerConfig {
   SherpaOnnxFeatureConfig feat_config;
   SherpaOnnxOfflineModelConfig model_config;
   SherpaOnnxOfflineLMConfig lm_config;
@@ -572,10 +572,10 @@ SHERPA_ONNX_API typedef struct SherpaOnnxOfflineRecognizerConfig {
   SherpaOnnxHomophoneReplacerConfig hr;
 } SherpaOnnxOfflineRecognizerConfig;
 
-SHERPA_ONNX_API typedef struct SherpaOnnxOfflineRecognizer
+typedef struct SherpaOnnxOfflineRecognizer
     SherpaOnnxOfflineRecognizer;
 
-SHERPA_ONNX_API typedef struct SherpaOnnxOfflineStream SherpaOnnxOfflineStream;
+typedef struct SherpaOnnxOfflineStream SherpaOnnxOfflineStream;
 
 /// @param config  Config for the recognizer.
 /// @return Return a pointer to the recognizer. The user has to invoke
@@ -660,7 +660,7 @@ SHERPA_ONNX_API void SherpaOnnxDecodeMultipleOfflineStreams(
     const SherpaOnnxOfflineRecognizer *recognizer,
     const SherpaOnnxOfflineStream **streams, int32_t n);
 
-SHERPA_ONNX_API typedef struct SherpaOnnxOfflineRecognizerResult {
+typedef struct SherpaOnnxOfflineRecognizerResult {
   const char *text;
 
   // Pointer to continuous memory which holds timestamps
@@ -761,7 +761,7 @@ SHERPA_ONNX_API void SherpaOnnxDestroyOfflineStreamResultJson(const char *s);
 // ============================================================
 // For Keyword Spotter
 // ============================================================
-SHERPA_ONNX_API typedef struct SherpaOnnxKeywordResult {
+typedef struct SherpaOnnxKeywordResult {
   /// The triggered keyword.
   /// For English, it consists of space separated words.
   /// For Chinese, it consists of Chinese words without spaces.
@@ -798,7 +798,7 @@ SHERPA_ONNX_API typedef struct SherpaOnnxKeywordResult {
   const char *json;
 } SherpaOnnxKeywordResult;
 
-SHERPA_ONNX_API typedef struct SherpaOnnxKeywordSpotterConfig {
+typedef struct SherpaOnnxKeywordSpotterConfig {
   SherpaOnnxFeatureConfig feat_config;
   SherpaOnnxOnlineModelConfig model_config;
   int32_t max_active_paths;
@@ -813,7 +813,7 @@ SHERPA_ONNX_API typedef struct SherpaOnnxKeywordSpotterConfig {
   int32_t keywords_buf_size;
 } SherpaOnnxKeywordSpotterConfig;
 
-SHERPA_ONNX_API typedef struct SherpaOnnxKeywordSpotter
+typedef struct SherpaOnnxKeywordSpotter
     SherpaOnnxKeywordSpotter;
 
 /// @param config  Config for the keyword spotter.
@@ -912,7 +912,7 @@ SHERPA_ONNX_API void SherpaOnnxFreeKeywordResultJson(const char *s);
 // For VAD
 // ============================================================
 
-SHERPA_ONNX_API typedef struct SherpaOnnxSileroVadModelConfig {
+typedef struct SherpaOnnxSileroVadModelConfig {
   // Path to the silero VAD model
   const char *model;
 
@@ -936,7 +936,7 @@ SHERPA_ONNX_API typedef struct SherpaOnnxSileroVadModelConfig {
   float max_speech_duration;
 } SherpaOnnxSileroVadModelConfig;
 
-SHERPA_ONNX_API typedef struct SherpaOnnxTenVadModelConfig {
+typedef struct SherpaOnnxTenVadModelConfig {
   // Path to the ten-vad model
   const char *model;
 
@@ -960,7 +960,7 @@ SHERPA_ONNX_API typedef struct SherpaOnnxTenVadModelConfig {
   float max_speech_duration;
 } SherpaOnnxTenVadModelConfig;
 
-SHERPA_ONNX_API typedef struct SherpaOnnxVadModelConfig {
+typedef struct SherpaOnnxVadModelConfig {
   SherpaOnnxSileroVadModelConfig silero_vad;
 
   int32_t sample_rate;
@@ -970,7 +970,7 @@ SHERPA_ONNX_API typedef struct SherpaOnnxVadModelConfig {
   SherpaOnnxTenVadModelConfig ten_vad;
 } SherpaOnnxVadModelConfig;
 
-SHERPA_ONNX_API typedef struct SherpaOnnxCircularBuffer
+typedef struct SherpaOnnxCircularBuffer
     SherpaOnnxCircularBuffer;
 
 // Return an instance of circular buffer. The user has to use
@@ -1014,7 +1014,7 @@ SherpaOnnxCircularBufferHead(const SherpaOnnxCircularBuffer *buffer);
 SHERPA_ONNX_API void SherpaOnnxCircularBufferReset(
     const SherpaOnnxCircularBuffer *buffer);
 
-SHERPA_ONNX_API typedef struct SherpaOnnxSpeechSegment {
+typedef struct SherpaOnnxSpeechSegment {
   // The start index in samples of this segment
   int32_t start;
 
@@ -1079,7 +1079,7 @@ SHERPA_ONNX_API void SherpaOnnxVoiceActivityDetectorFlush(
 // ============================================================
 // For offline Text-to-Speech (i.e., non-streaming TTS)
 // ============================================================
-SHERPA_ONNX_API typedef struct SherpaOnnxOfflineTtsVitsModelConfig {
+typedef struct SherpaOnnxOfflineTtsVitsModelConfig {
   const char *model;
   const char *lexicon;
   const char *tokens;
@@ -1091,7 +1091,7 @@ SHERPA_ONNX_API typedef struct SherpaOnnxOfflineTtsVitsModelConfig {
   const char *dict_dir;  // unused
 } SherpaOnnxOfflineTtsVitsModelConfig;
 
-SHERPA_ONNX_API typedef struct SherpaOnnxOfflineTtsMatchaModelConfig {
+typedef struct SherpaOnnxOfflineTtsMatchaModelConfig {
   const char *acoustic_model;
   const char *vocoder;
   const char *lexicon;
@@ -1103,7 +1103,7 @@ SHERPA_ONNX_API typedef struct SherpaOnnxOfflineTtsMatchaModelConfig {
   const char *dict_dir;  // unused
 } SherpaOnnxOfflineTtsMatchaModelConfig;
 
-SHERPA_ONNX_API typedef struct SherpaOnnxOfflineTtsKokoroModelConfig {
+typedef struct SherpaOnnxOfflineTtsKokoroModelConfig {
   const char *model;
   const char *voices;
   const char *tokens;
@@ -1115,7 +1115,7 @@ SHERPA_ONNX_API typedef struct SherpaOnnxOfflineTtsKokoroModelConfig {
   const char *lang;
 } SherpaOnnxOfflineTtsKokoroModelConfig;
 
-SHERPA_ONNX_API typedef struct SherpaOnnxOfflineTtsKittenModelConfig {
+typedef struct SherpaOnnxOfflineTtsKittenModelConfig {
   const char *model;
   const char *voices;
   const char *tokens;
@@ -1124,7 +1124,7 @@ SHERPA_ONNX_API typedef struct SherpaOnnxOfflineTtsKittenModelConfig {
   float length_scale;  // < 1, faster in speech speed; > 1, slower in speed
 } SherpaOnnxOfflineTtsKittenModelConfig;
 
-SHERPA_ONNX_API typedef struct SherpaOnnxOfflineTtsZipvoiceModelConfig {
+typedef struct SherpaOnnxOfflineTtsZipvoiceModelConfig {
   const char *tokens;
   const char *encoder;
   const char *decoder;
@@ -1137,7 +1137,7 @@ SHERPA_ONNX_API typedef struct SherpaOnnxOfflineTtsZipvoiceModelConfig {
   float guidance_scale;
 } SherpaOnnxOfflineTtsZipvoiceModelConfig;
 
-SHERPA_ONNX_API typedef struct SherpaOnnxOfflineTtsPocketModelConfig {
+typedef struct SherpaOnnxOfflineTtsPocketModelConfig {
   const char *lm_flow;
   const char *lm_main;
   const char *encoder;
@@ -1148,7 +1148,7 @@ SHERPA_ONNX_API typedef struct SherpaOnnxOfflineTtsPocketModelConfig {
   int32_t voice_embedding_cache_capacity;
 } SherpaOnnxOfflineTtsPocketModelConfig;
 
-SHERPA_ONNX_API typedef struct SherpaOnnxOfflineTtsSupertonicModelConfig {
+typedef struct SherpaOnnxOfflineTtsSupertonicModelConfig {
   const char *duration_predictor;
   const char *text_encoder;
   const char *vector_estimator;
@@ -1158,7 +1158,7 @@ SHERPA_ONNX_API typedef struct SherpaOnnxOfflineTtsSupertonicModelConfig {
   const char *voice_style;
 } SherpaOnnxOfflineTtsSupertonicModelConfig;
 
-SHERPA_ONNX_API typedef struct SherpaOnnxOfflineTtsModelConfig {
+typedef struct SherpaOnnxOfflineTtsModelConfig {
   SherpaOnnxOfflineTtsVitsModelConfig vits;
   int32_t num_threads;
   int32_t debug;
@@ -1171,7 +1171,7 @@ SHERPA_ONNX_API typedef struct SherpaOnnxOfflineTtsModelConfig {
   SherpaOnnxOfflineTtsSupertonicModelConfig supertonic;
 } SherpaOnnxOfflineTtsModelConfig;
 
-SHERPA_ONNX_API typedef struct SherpaOnnxOfflineTtsConfig {
+typedef struct SherpaOnnxOfflineTtsConfig {
   SherpaOnnxOfflineTtsModelConfig model;
   const char *rule_fsts;
   int32_t max_num_sentences;
@@ -1179,7 +1179,7 @@ SHERPA_ONNX_API typedef struct SherpaOnnxOfflineTtsConfig {
   float silence_scale;
 } SherpaOnnxOfflineTtsConfig;
 
-SHERPA_ONNX_API typedef struct SherpaOnnxGeneratedAudio {
+typedef struct SherpaOnnxGeneratedAudio {
   const float *samples;  // in the range [-1, 1]
   int32_t n;             // number of samples
   int32_t sample_rate;
@@ -1200,7 +1200,7 @@ typedef int32_t (*SherpaOnnxGeneratedAudioProgressCallback)(
 typedef int32_t (*SherpaOnnxGeneratedAudioProgressCallbackWithArg)(
     const float *samples, int32_t n, float p, void *arg);
 
-SHERPA_ONNX_API typedef struct SherpaOnnxOfflineTts SherpaOnnxOfflineTts;
+typedef struct SherpaOnnxOfflineTts SherpaOnnxOfflineTts;
 
 // Create an instance of offline TTS. The user has to use DestroyOfflineTts()
 // to free the returned pointer to avoid memory leak.
@@ -1262,7 +1262,7 @@ SherpaOnnxOfflineTtsGenerateWithZipvoice(const SherpaOnnxOfflineTts *tts,
                                          int32_t n_prompt, int32_t prompt_sr,
                                          float speed, int32_t num_steps);
 
-SHERPA_ONNX_API typedef struct SherpaOnnxGenerationConfig {
+typedef struct SherpaOnnxGenerationConfig {
   float silence_scale;
   float speed;                    // used only by some models.
   int32_t sid;                    // used only by models support multi-speakers
@@ -1306,7 +1306,7 @@ SHERPA_ONNX_API void SherpaOnnxWriteWaveToBuffer(const float *samples,
                                                  int32_t n, int32_t sample_rate,
                                                  char *buffer);
 
-SHERPA_ONNX_API typedef struct SherpaOnnxWave {
+typedef struct SherpaOnnxWave {
   // samples normalized to the range [-1, 1]
   const float *samples;
   int32_t sample_rate;
@@ -1334,21 +1334,21 @@ SHERPA_ONNX_API void SherpaOnnxFreeWave(const SherpaOnnxWave *wave);
 // For spoken language identification
 // ============================================================
 
-SHERPA_ONNX_API typedef struct
+typedef struct
     SherpaOnnxSpokenLanguageIdentificationWhisperConfig {
   const char *encoder;
   const char *decoder;
   int32_t tail_paddings;
 } SherpaOnnxSpokenLanguageIdentificationWhisperConfig;
 
-SHERPA_ONNX_API typedef struct SherpaOnnxSpokenLanguageIdentificationConfig {
+typedef struct SherpaOnnxSpokenLanguageIdentificationConfig {
   SherpaOnnxSpokenLanguageIdentificationWhisperConfig whisper;
   int32_t num_threads;
   int32_t debug;
   const char *provider;
 } SherpaOnnxSpokenLanguageIdentificationConfig;
 
-SHERPA_ONNX_API typedef struct SherpaOnnxSpokenLanguageIdentification
+typedef struct SherpaOnnxSpokenLanguageIdentification
     SherpaOnnxSpokenLanguageIdentification;
 
 // Create an instance of SpokenLanguageIdentification.
@@ -1367,7 +1367,7 @@ SHERPA_ONNX_API SherpaOnnxOfflineStream *
 SherpaOnnxSpokenLanguageIdentificationCreateOfflineStream(
     const SherpaOnnxSpokenLanguageIdentification *slid);
 
-SHERPA_ONNX_API typedef struct SherpaOnnxSpokenLanguageIdentificationResult {
+typedef struct SherpaOnnxSpokenLanguageIdentificationResult {
   // en for English
   // de for German
   // zh for Chinese
@@ -1389,14 +1389,14 @@ SHERPA_ONNX_API void SherpaOnnxDestroySpokenLanguageIdentificationResult(
 // ============================================================
 // For speaker embedding extraction
 // ============================================================
-SHERPA_ONNX_API typedef struct SherpaOnnxSpeakerEmbeddingExtractorConfig {
+typedef struct SherpaOnnxSpeakerEmbeddingExtractorConfig {
   const char *model;
   int32_t num_threads;
   int32_t debug;
   const char *provider;
 } SherpaOnnxSpeakerEmbeddingExtractorConfig;
 
-SHERPA_ONNX_API typedef struct SherpaOnnxSpeakerEmbeddingExtractor
+typedef struct SherpaOnnxSpeakerEmbeddingExtractor
     SherpaOnnxSpeakerEmbeddingExtractor;
 
 // The user has to invoke SherpaOnnxDestroySpeakerEmbeddingExtractor()
@@ -1439,7 +1439,7 @@ SherpaOnnxSpeakerEmbeddingExtractorComputeEmbedding(
 SHERPA_ONNX_API void SherpaOnnxSpeakerEmbeddingExtractorDestroyEmbedding(
     const float *v);
 
-SHERPA_ONNX_API typedef struct SherpaOnnxSpeakerEmbeddingManager
+typedef struct SherpaOnnxSpeakerEmbeddingManager
     SherpaOnnxSpeakerEmbeddingManager;
 
 // The user has to invoke SherpaOnnxDestroySpeakerEmbeddingManager()
@@ -1506,12 +1506,12 @@ SHERPA_ONNX_API const char *SherpaOnnxSpeakerEmbeddingManagerSearch(
 SHERPA_ONNX_API void SherpaOnnxSpeakerEmbeddingManagerFreeSearch(
     const char *name);
 
-SHERPA_ONNX_API typedef struct SherpaOnnxSpeakerEmbeddingManagerSpeakerMatch {
+typedef struct SherpaOnnxSpeakerEmbeddingManagerSpeakerMatch {
   float score;
   const char *name;
 } SherpaOnnxSpeakerEmbeddingManagerSpeakerMatch;
 
-SHERPA_ONNX_API typedef struct
+typedef struct
     SherpaOnnxSpeakerEmbeddingManagerBestMatchesResult {
   const SherpaOnnxSpeakerEmbeddingManagerSpeakerMatch *matches;
   int32_t count;
@@ -1585,12 +1585,12 @@ SHERPA_ONNX_API void SherpaOnnxSpeakerEmbeddingManagerFreeAllSpeakers(
 // ============================================================
 // For audio tagging
 // ============================================================
-SHERPA_ONNX_API typedef struct
+typedef struct
     SherpaOnnxOfflineZipformerAudioTaggingModelConfig {
   const char *model;
 } SherpaOnnxOfflineZipformerAudioTaggingModelConfig;
 
-SHERPA_ONNX_API typedef struct SherpaOnnxAudioTaggingModelConfig {
+typedef struct SherpaOnnxAudioTaggingModelConfig {
   SherpaOnnxOfflineZipformerAudioTaggingModelConfig zipformer;
   const char *ced;
   int32_t num_threads;
@@ -1598,19 +1598,19 @@ SHERPA_ONNX_API typedef struct SherpaOnnxAudioTaggingModelConfig {
   const char *provider;
 } SherpaOnnxAudioTaggingModelConfig;
 
-SHERPA_ONNX_API typedef struct SherpaOnnxAudioTaggingConfig {
+typedef struct SherpaOnnxAudioTaggingConfig {
   SherpaOnnxAudioTaggingModelConfig model;
   const char *labels;
   int32_t top_k;
 } SherpaOnnxAudioTaggingConfig;
 
-SHERPA_ONNX_API typedef struct SherpaOnnxAudioEvent {
+typedef struct SherpaOnnxAudioEvent {
   const char *name;
   int32_t index;
   float prob;
 } SherpaOnnxAudioEvent;
 
-SHERPA_ONNX_API typedef struct SherpaOnnxAudioTagging SherpaOnnxAudioTagging;
+typedef struct SherpaOnnxAudioTagging SherpaOnnxAudioTagging;
 
 // The user has to invoke
 // SherpaOnnxDestroyAudioTagging()
@@ -1646,18 +1646,18 @@ SHERPA_ONNX_API void SherpaOnnxAudioTaggingFreeResults(
 // For punctuation
 // ============================================================
 
-SHERPA_ONNX_API typedef struct SherpaOnnxOfflinePunctuationModelConfig {
+typedef struct SherpaOnnxOfflinePunctuationModelConfig {
   const char *ct_transformer;
   int32_t num_threads;
   int32_t debug;  // true to print debug information of the model
   const char *provider;
 } SherpaOnnxOfflinePunctuationModelConfig;
 
-SHERPA_ONNX_API typedef struct SherpaOnnxOfflinePunctuationConfig {
+typedef struct SherpaOnnxOfflinePunctuationConfig {
   SherpaOnnxOfflinePunctuationModelConfig model;
 } SherpaOnnxOfflinePunctuationConfig;
 
-SHERPA_ONNX_API typedef struct SherpaOnnxOfflinePunctuation
+typedef struct SherpaOnnxOfflinePunctuation
     SherpaOnnxOfflinePunctuation;
 
 // The user has to invoke SherpaOnnxDestroyOfflinePunctuation()
@@ -1677,7 +1677,7 @@ SHERPA_ONNX_API const char *SherpaOfflinePunctuationAddPunct(
 
 SHERPA_ONNX_API void SherpaOfflinePunctuationFreeText(const char *text);
 
-SHERPA_ONNX_API typedef struct SherpaOnnxOnlinePunctuationModelConfig {
+typedef struct SherpaOnnxOnlinePunctuationModelConfig {
   const char *cnn_bilstm;
   const char *bpe_vocab;
   int32_t num_threads;
@@ -1685,11 +1685,11 @@ SHERPA_ONNX_API typedef struct SherpaOnnxOnlinePunctuationModelConfig {
   const char *provider;
 } SherpaOnnxOnlinePunctuationModelConfig;
 
-SHERPA_ONNX_API typedef struct SherpaOnnxOnlinePunctuationConfig {
+typedef struct SherpaOnnxOnlinePunctuationConfig {
   SherpaOnnxOnlinePunctuationModelConfig model;
 } SherpaOnnxOnlinePunctuationConfig;
 
-SHERPA_ONNX_API typedef struct SherpaOnnxOnlinePunctuation
+typedef struct SherpaOnnxOnlinePunctuation
     SherpaOnnxOnlinePunctuation;
 
 // Create an online punctuation processor. The user has to invoke
@@ -1713,7 +1713,7 @@ SHERPA_ONNX_API const char *SherpaOnnxOnlinePunctuationAddPunct(
 SHERPA_ONNX_API void SherpaOnnxOnlinePunctuationFreeText(const char *text);
 
 // for resampling
-SHERPA_ONNX_API typedef struct SherpaOnnxLinearResampler
+typedef struct SherpaOnnxLinearResampler
     SherpaOnnxLinearResampler;
 
 /*
@@ -1762,19 +1762,19 @@ SHERPA_ONNX_API int32_t SherpaOnnxLinearResamplerResampleGetOutputSampleRate(
 // =========================================================================
 // For offline speaker diarization (i.e., non-streaming speaker diarization)
 // =========================================================================
-SHERPA_ONNX_API typedef struct
+typedef struct
     SherpaOnnxOfflineSpeakerSegmentationPyannoteModelConfig {
   const char *model;
 } SherpaOnnxOfflineSpeakerSegmentationPyannoteModelConfig;
 
-SHERPA_ONNX_API typedef struct SherpaOnnxOfflineSpeakerSegmentationModelConfig {
+typedef struct SherpaOnnxOfflineSpeakerSegmentationModelConfig {
   SherpaOnnxOfflineSpeakerSegmentationPyannoteModelConfig pyannote;
   int32_t num_threads;   // 1
   int32_t debug;         // false
   const char *provider;  // "cpu"
 } SherpaOnnxOfflineSpeakerSegmentationModelConfig;
 
-SHERPA_ONNX_API typedef struct SherpaOnnxFastClusteringConfig {
+typedef struct SherpaOnnxFastClusteringConfig {
   // If greater than 0, then threshold is ignored.
   //
   // We strongly recommend that you set it if you know the number of clusters
@@ -1788,7 +1788,7 @@ SHERPA_ONNX_API typedef struct SherpaOnnxFastClusteringConfig {
   float threshold;
 } SherpaOnnxFastClusteringConfig;
 
-SHERPA_ONNX_API typedef struct SherpaOnnxOfflineSpeakerDiarizationConfig {
+typedef struct SherpaOnnxOfflineSpeakerDiarizationConfig {
   SherpaOnnxOfflineSpeakerSegmentationModelConfig segmentation;
   SherpaOnnxSpeakerEmbeddingExtractorConfig embedding;
   SherpaOnnxFastClusteringConfig clustering;
@@ -1802,7 +1802,7 @@ SHERPA_ONNX_API typedef struct SherpaOnnxOfflineSpeakerDiarizationConfig {
   float min_duration_off;  // in seconds
 } SherpaOnnxOfflineSpeakerDiarizationConfig;
 
-SHERPA_ONNX_API typedef struct SherpaOnnxOfflineSpeakerDiarization
+typedef struct SherpaOnnxOfflineSpeakerDiarization
     SherpaOnnxOfflineSpeakerDiarization;
 
 // The users has to invoke SherpaOnnxDestroyOfflineSpeakerDiarization()
@@ -1824,10 +1824,10 @@ SHERPA_ONNX_API void SherpaOnnxOfflineSpeakerDiarizationSetConfig(
     const SherpaOnnxOfflineSpeakerDiarization *sd,
     const SherpaOnnxOfflineSpeakerDiarizationConfig *config);
 
-SHERPA_ONNX_API typedef struct SherpaOnnxOfflineSpeakerDiarizationResult
+typedef struct SherpaOnnxOfflineSpeakerDiarizationResult
     SherpaOnnxOfflineSpeakerDiarizationResult;
 
-SHERPA_ONNX_API typedef struct SherpaOnnxOfflineSpeakerDiarizationSegment {
+typedef struct SherpaOnnxOfflineSpeakerDiarizationSegment {
   float start;
   float end;
   int32_t speaker;
@@ -1885,22 +1885,22 @@ SHERPA_ONNX_API void SherpaOnnxOfflineSpeakerDiarizationDestroyResult(
 // =========================================================================
 // For offline speech enhancement
 // =========================================================================
-SHERPA_ONNX_API typedef struct SherpaOnnxOfflineSpeechDenoiserGtcrnModelConfig {
+typedef struct SherpaOnnxOfflineSpeechDenoiserGtcrnModelConfig {
   const char *model;
 } SherpaOnnxOfflineSpeechDenoiserGtcrnModelConfig;
 
-SHERPA_ONNX_API typedef struct SherpaOnnxOfflineSpeechDenoiserModelConfig {
+typedef struct SherpaOnnxOfflineSpeechDenoiserModelConfig {
   SherpaOnnxOfflineSpeechDenoiserGtcrnModelConfig gtcrn;
   int32_t num_threads;
   int32_t debug;  // true to print debug information of the model
   const char *provider;
 } SherpaOnnxOfflineSpeechDenoiserModelConfig;
 
-SHERPA_ONNX_API typedef struct SherpaOnnxOfflineSpeechDenoiserConfig {
+typedef struct SherpaOnnxOfflineSpeechDenoiserConfig {
   SherpaOnnxOfflineSpeechDenoiserModelConfig model;
 } SherpaOnnxOfflineSpeechDenoiserConfig;
 
-SHERPA_ONNX_API typedef struct SherpaOnnxOfflineSpeechDenoiser
+typedef struct SherpaOnnxOfflineSpeechDenoiser
     SherpaOnnxOfflineSpeechDenoiser;
 
 // The users has to invoke SherpaOnnxDestroyOfflineSpeechDenoiser()
@@ -1916,7 +1916,7 @@ SHERPA_ONNX_API void SherpaOnnxDestroyOfflineSpeechDenoiser(
 SHERPA_ONNX_API int32_t SherpaOnnxOfflineSpeechDenoiserGetSampleRate(
     const SherpaOnnxOfflineSpeechDenoiser *sd);
 
-SHERPA_ONNX_API typedef struct SherpaOnnxDenoisedAudio {
+typedef struct SherpaOnnxDenoisedAudio {
   const float *samples;  // in the range [-1, 1]
   int32_t n;             // number of samples
   int32_t sample_rate;
diff --git a/sherpa-onnx/c-api/cxx-api.h b/sherpa-onnx/c-api/cxx-api.h
index 8d4af2d3ea..359bbf9b3a 100644
--- a/sherpa-onnx/c-api/cxx-api.h
+++ b/sherpa-onnx/c-api/cxx-api.h
@@ -212,21 +212,21 @@ class SHERPA_ONNX_API OnlineRecognizer
 // ============================================================================
 // Non-streaming ASR
 // ============================================================================
-struct SHERPA_ONNX_API OfflineTransducerModelConfig {
+struct OfflineTransducerModelConfig {
   std::string encoder;
   std::string decoder;
   std::string joiner;
 };
 
-struct SHERPA_ONNX_API OfflineParaformerModelConfig {
+struct OfflineParaformerModelConfig {
   std::string model;
 };
 
-struct SHERPA_ONNX_API OfflineNemoEncDecCtcModelConfig {
+struct OfflineNemoEncDecCtcModelConfig {
   std::string model;
 };
 
-struct SHERPA_ONNX_API OfflineWhisperModelConfig {
+struct OfflineWhisperModelConfig {
   std::string encoder;
   std::string decoder;
   std::string language;
@@ -234,7 +234,7 @@ struct SHERPA_ONNX_API OfflineWhisperModelConfig {
   int32_t tail_paddings = -1;
 };
 
-struct SHERPA_ONNX_API OfflineCanaryModelConfig {
+struct OfflineCanaryModelConfig {
   std::string encoder;
   std::string decoder;
   std::string src_lang;
@@ -242,46 +242,46 @@ struct SHERPA_ONNX_API OfflineCanaryModelConfig {
   bool use_pnc = true;
 };
 
-struct SHERPA_ONNX_API OfflineFireRedAsrModelConfig {
+struct OfflineFireRedAsrModelConfig {
   std::string encoder;
   std::string decoder;
 };
 
-struct SHERPA_ONNX_API OfflineFireRedAsrCtcModelConfig {
+struct OfflineFireRedAsrCtcModelConfig {
   std::string model;
 };
 
-struct SHERPA_ONNX_API OfflineTdnnModelConfig {
+struct OfflineTdnnModelConfig {
   std::string model;
 };
 
-struct SHERPA_ONNX_API OfflineSenseVoiceModelConfig {
+struct OfflineSenseVoiceModelConfig {
   std::string model;
   std::string language;
   bool use_itn = false;
 };
 
-struct SHERPA_ONNX_API OfflineDolphinModelConfig {
+struct OfflineDolphinModelConfig {
   std::string model;
 };
 
-struct SHERPA_ONNX_API OfflineZipformerCtcModelConfig {
+struct OfflineZipformerCtcModelConfig {
   std::string model;
 };
 
-struct SHERPA_ONNX_API OfflineWenetCtcModelConfig {
+struct OfflineWenetCtcModelConfig {
   std::string model;
 };
 
-struct SHERPA_ONNX_API OfflineOmnilingualAsrCtcModelConfig {
+struct OfflineOmnilingualAsrCtcModelConfig {
   std::string model;
 };
 
-struct SHERPA_ONNX_API OfflineMedAsrCtcModelConfig {
+struct OfflineMedAsrCtcModelConfig {
   std::string model;
 };
 
-struct SHERPA_ONNX_API OfflineMoonshineModelConfig {
+struct OfflineMoonshineModelConfig {
   std::string preprocessor;
   std::string encoder;
   std::string uncached_decoder;
@@ -289,7 +289,7 @@ struct SHERPA_ONNX_API OfflineMoonshineModelConfig {
   std::string merged_decoder;
 };
 
-struct SHERPA_ONNX_API OfflineFunASRNanoModelConfig {
+struct OfflineFunASRNanoModelConfig {
   std::string encoder_adaptor;
   std::string llm;
   std::string embedding;
@@ -305,7 +305,7 @@ struct SHERPA_ONNX_API OfflineFunASRNanoModelConfig {
   std::string hotwords;
 };
 
-struct SHERPA_ONNX_API OfflineModelConfig {
+struct OfflineModelConfig {
   OfflineTransducerModelConfig transducer;
   OfflineParaformerModelConfig paraformer;
   OfflineNemoEncDecCtcModelConfig nemo_ctc;
@@ -333,12 +333,12 @@ struct SHERPA_ONNX_API OfflineModelConfig {
   OfflineFireRedAsrCtcModelConfig fire_red_asr_ctc;
 };
 
-struct SHERPA_ONNX_API OfflineLMConfig {
+struct OfflineLMConfig {
   std::string model;
   float scale = 1.0;
 };
 
-struct SHERPA_ONNX_API OfflineRecognizerConfig {
+struct OfflineRecognizerConfig {
   FeatureConfig feat_config;
   OfflineModelConfig model_config;
   OfflineLMConfig lm_config;
@@ -355,7 +355,7 @@ struct SHERPA_ONNX_API OfflineRecognizerConfig {
   HomophoneReplacerConfig hr;
 };
 
-struct SHERPA_ONNX_API OfflineRecognizerResult {
+struct OfflineRecognizerResult {
   std::string text;
   std::vector<float> timestamps;
   std::vector<std::string> tokens;