From 7931f6520db649049343d165d2734dddb49e406b Mon Sep 17 00:00:00 2001
From: Fangjun Kuang <csukuangfj@gmail.com>
Date: Thu, 18 Jul 2024 21:15:14 +0800
Subject: [PATCH] swift api for sense voice

---
 .gitignore                                    |  1 +
 swift-api-examples/SherpaOnnx.swift           | 18 ++++++++++++++++--
 .../decode-file-non-streaming.swift           | 19 ++++++++++++++++++-
 3 files changed, 35 insertions(+), 3 deletions(-)

diff --git a/.gitignore b/.gitignore
index a39684cee..5486ad51a 100644
--- a/.gitignore
+++ b/.gitignore
@@ -111,3 +111,4 @@ sherpa-onnx-telespeech-ctc-*
 *.fst
 .ccache
 lib*.a
+sherpa-onnx-sense-voice-zh-en-ja-ko-yue-2024-07-17
diff --git a/swift-api-examples/SherpaOnnx.swift b/swift-api-examples/SherpaOnnx.swift
index eba8d8916..c6f8b51eb 100644
--- a/swift-api-examples/SherpaOnnx.swift
+++ b/swift-api-examples/SherpaOnnx.swift
@@ -355,6 +355,18 @@ func sherpaOnnxOfflineTdnnModelConfig(
   )
 }
 
+func sherpaOnnxOfflineSenseVoiceModelConfig(
+  model: String = "",
+  language: String = "",
+  useInverseTextNormalization: Bool = false
+) -> SherpaOnnxOfflineSenseVoiceModelConfig {
+  return SherpaOnnxOfflineSenseVoiceModelConfig(
+    model: toCPointer(model),
+    language: toCPointer(language),
+    use_itn: useInverseTextNormalization ? 1 : 0
+  )
+}
+
 func sherpaOnnxOfflineLMConfig(
   model: String = "",
   scale: Float = 1.0
@@ -378,7 +390,8 @@ func sherpaOnnxOfflineModelConfig(
   modelType: String = "",
   modelingUnit: String = "cjkchar",
   bpeVocab: String = "",
-  teleSpeechCtc: String = ""
+  teleSpeechCtc: String = "",
+  senseVoice: SherpaOnnxOfflineSenseVoiceModelConfig = sherpaOnnxOfflineSenseVoiceModelConfig()
 ) -> SherpaOnnxOfflineModelConfig {
   return SherpaOnnxOfflineModelConfig(
     transducer: transducer,
@@ -393,7 +406,8 @@ func sherpaOnnxOfflineModelConfig(
     model_type: toCPointer(modelType),
     modeling_unit: toCPointer(modelingUnit),
     bpe_vocab: toCPointer(bpeVocab),
-    telespeech_ctc: toCPointer(teleSpeechCtc)
+    telespeech_ctc: toCPointer(teleSpeechCtc),
+    sense_voice: senseVoice
   )
 }
 
diff --git a/swift-api-examples/decode-file-non-streaming.swift b/swift-api-examples/decode-file-non-streaming.swift
index ca9d9475e..a60777832 100644
--- a/swift-api-examples/decode-file-non-streaming.swift
+++ b/swift-api-examples/decode-file-non-streaming.swift
@@ -17,6 +17,7 @@ func run() {
   var modelConfig: SherpaOnnxOfflineModelConfig
   var modelType = "whisper"
   // modelType = "paraformer"
+  // modelType = "sense_voice"
 
   if modelType == "whisper" {
     let encoder = "./sherpa-onnx-whisper-tiny.en/tiny.en-encoder.int8.onnx"
@@ -47,6 +48,19 @@ func run() {
       debug: 0,
       modelType: "paraformer"
     )
+  } else if modelType == "sense_voice" {
+    let model = "./sherpa-onnx-sense-voice-zh-en-ja-ko-yue-2024-07-17/model.int8.onnx"
+    let tokens = "./sherpa-onnx-sense-voice-zh-en-ja-ko-yue-2024-07-17/tokens.txt"
+    let senseVoiceConfig = sherpaOnnxOfflineSenseVoiceModelConfig(
+      model: model,
+      useInverseTextNormalization: true
+    )
+
+    modelConfig = sherpaOnnxOfflineModelConfig(
+      tokens: tokens,
+      debug: 0,
+      senseVoice: senseVoiceConfig
+    )
   } else {
     print("Please specify a supported modelType \(modelType)")
     return
@@ -63,7 +77,10 @@ func run() {
 
   recognizer = SherpaOnnxOfflineRecognizer(config: &config)
 
-  let filePath = "./sherpa-onnx-whisper-tiny.en/test_wavs/0.wav"
+  var filePath = "./sherpa-onnx-whisper-tiny.en/test_wavs/0.wav"
+  if modelType == "sense_voice" {
+    filePath = "./sherpa-onnx-sense-voice-zh-en-ja-ko-yue-2024-07-17/test_wavs/zh.wav"
+  }
   let fileURL: NSURL = NSURL(fileURLWithPath: filePath)
   let audioFile = try! AVAudioFile(forReading: fileURL as URL)