Before consolidated main

src/cpp/main.cpp

@@ -188,7 +188,7 @@ void parseArgs(int argc, char *argv[], RunConfig &runConfig) {
       runConfig.outputPath = filesystem::path(argv[++i]);
     } else if (arg == "-s" || arg == "--speaker") {
       ensureArg(argc, argv, i);
-      runConfig.speakerId = (larynx::SpeakerId)stoi(argv[++i]);
+      runConfig.speakerId = (larynx::SpeakerId)stoll(argv[++i]);
     } else if (arg == "-h" || arg == "--help") {
       printUsage(argv);
       exit(0);

src/cpp/model.hpp

@@ -16,15 +16,16 @@ struct ModelSession {
   vector<char *> outputNames;
   Ort::AllocatorWithDefaultOptions allocator;
   Ort::SessionOptions options;
+  Ort::Env env;
 
   ModelSession() : onnx(nullptr){};
 };
 
 void loadModel(string modelPath, ModelSession &session) {
 
-  Ort::Env env(OrtLoggingLevel::ORT_LOGGING_LEVEL_WARNING,
-               instanceName.c_str());
-  env.DisableTelemetryEvents();
+  session.env = Ort::Env(OrtLoggingLevel::ORT_LOGGING_LEVEL_WARNING,
+                         instanceName.c_str());
+  session.env.DisableTelemetryEvents();
 
   // Slows down performance by ~2x
   // session.options.SetIntraOpNumThreads(1);
@@ -44,7 +45,7 @@ void loadModel(string modelPath, ModelSession &session) {
   session.options.DisableProfiling();
 
   auto startTime = chrono::steady_clock::now();
-  session.onnx = Ort::Session(env, modelPath.c_str(), session.options);
+  session.onnx = Ort::Session(session.env, modelPath.c_str(), session.options);
   auto endTime = chrono::steady_clock::now();
   auto loadDuration = chrono::duration<double>(endTime - startTime);
 

src/cpp/synthesize.hpp

@@ -56,7 +56,7 @@ void synthesize(SynthesisConfig &synthesisConfig, ModelSession &session,
   if (synthesisConfig.speakerId) {
     // Add speaker id
     vector<int64_t> speakerId{(int64_t)synthesisConfig.speakerId.value()};
-    vector<int64_t> speakerIdShape{1};
+    vector<int64_t> speakerIdShape{(int64_t)speakerId.size()};
     inputTensors.push_back(Ort::Value::CreateTensor<int64_t>(
         memoryInfo, speakerId.data(), speakerId.size(), speakerIdShape.data(),
         speakerIdShape.size()));