speech_recognition/FunASR
1
0
Fork 0
mirror of https://github.com/modelscope/FunASR synced 2025-09-15 14:48:36 +08:00
Code Issues Actions Packages Projects Releases Wiki Activity

add GetInputNames GetOutputNames

Browse Source
This commit is contained in:
雾聪 2024-10-10 17:45:45 +08:00
parent bef2d3a391
commit 1480dcf5d5
7 changed files with 58 additions and 115 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

26
runtime/onnxruntime/src/commonfunc.h
View File

@ -65,6 +65,19 @@ inline void GetInputName(Ort::Session* session, string& inputName,int nIndex=0)
}
}
inline void GetInputNames(Ort::Session* session, std::vector<std::string> &m_strInputNames,
std::vector<const char *> &m_szInputNames) {
Ort::AllocatorWithDefaultOptions allocator;
size_t numNodes = session->GetInputCount();
m_strInputNames.resize(numNodes);
m_szInputNames.resize(numNodes);
for (size_t i = 0; i != numNodes; ++i) {
auto t = session->GetInputNameAllocated(i, allocator);
m_strInputNames[i] = t.get();
m_szInputNames[i] = m_strInputNames[i].c_str();
}
}
inline void GetOutputName(Ort::Session* session, string& outputName, int nIndex = 0) {
size_t numOutputNodes = session->GetOutputCount();
if (numOutputNodes > 0) {
@ -76,6 +89,19 @@ inline void GetOutputName(Ort::Session* session, string& outputName, int nIndex
}
}
inline void GetOutputNames(Ort::Session* session, std::vector<std::string> &m_strOutputNames,
std::vector<const char *> &m_szOutputNames) {
Ort::AllocatorWithDefaultOptions allocator;
size_t numNodes = session->GetOutputCount();
m_strOutputNames.resize(numNodes);
m_szOutputNames.resize(numNodes);
for (size_t i = 0; i != numNodes; ++i) {
auto t = session->GetOutputNameAllocated(i, allocator);
m_strOutputNames[i] = t.get();
m_szOutputNames[i] = m_strOutputNames[i].c_str();
}
}
template <class ForwardIterator>
inline static size_t Argmax(ForwardIterator first, ForwardIterator last) {
return std::distance(first, std::max_element(first, last));

19
runtime/onnxruntime/src/ct-transformer-online.cpp
View File

@ -25,23 +25,8 @@ void CTTransformerOnline::InitPunc(const std::string &punc_model, const std::str
exit(-1);
}
// read inputnames outputnames
string strName;
GetInputName(m_session.get(), strName);
m_strInputNames.push_back(strName.c_str());
GetInputName(m_session.get(), strName, 1);
m_strInputNames.push_back(strName);
GetInputName(m_session.get(), strName, 2);
m_strInputNames.push_back(strName);
GetInputName(m_session.get(), strName, 3);
m_strInputNames.push_back(strName);
GetOutputName(m_session.get(), strName);
m_strOutputNames.push_back(strName);
for (auto& item : m_strInputNames)
m_szInputNames.push_back(item.c_str());
for (auto& item : m_strOutputNames)
m_szOutputNames.push_back(item.c_str());
GetInputNames(m_session.get(), m_strInputNames, m_szInputNames);
GetOutputNames(m_session.get(), m_strOutputNames, m_szOutputNames);
m_tokenizer.OpenYaml(punc_config.c_str(), token_file.c_str());
}

15
runtime/onnxruntime/src/ct-transformer.cpp
View File

@ -25,20 +25,9 @@ void CTTransformer::InitPunc(const std::string &punc_model, const std::string &p
exit(-1);
}
// read inputnames outputnames
string strName;
GetInputName(m_session.get(), strName);
m_strInputNames.push_back(strName.c_str());
GetInputName(m_session.get(), strName, 1);
m_strInputNames.push_back(strName);
GetInputNames(m_session.get(), m_strInputNames, m_szInputNames);
GetOutputNames(m_session.get(), m_strOutputNames, m_szOutputNames);
GetOutputName(m_session.get(), strName);
m_strOutputNames.push_back(strName);
for (auto& item : m_strInputNames)
m_szInputNames.push_back(item.c_str());
for (auto& item : m_strOutputNames)
m_szOutputNames.push_back(item.c_str());
m_tokenizer.OpenYaml(punc_config.c_str(), token_file.c_str());
m_tokenizer.JiebaInit(punc_config);
}

24
runtime/onnxruntime/src/fsmn-vad.cpp
View File

@ -60,30 +60,10 @@ void FsmnVad::ReadModel(const char* vad_model) {
LOG(ERROR) << "Error when load vad onnx model: " << e.what();
exit(-1);
}
GetInputOutputInfo(vad_session_, &vad_in_names_, &vad_out_names_);
GetInputNames(vad_session_.get(), m_strInputNames, vad_in_names_);
GetOutputNames(vad_session_.get(), m_strOutputNames, vad_out_names_);
}
void FsmnVad::GetInputOutputInfo(
const std::shared_ptr<Ort::Session> &session,
std::vector<const char *> *in_names, std::vector<const char *> *out_names) {
Ort::AllocatorWithDefaultOptions allocator;
// Input info
int num_nodes = session->GetInputCount();
in_names->resize(num_nodes);
for (int i = 0; i < num_nodes; ++i) {
std::unique_ptr<char, Ort::detail::AllocatedFree> name = session->GetInputNameAllocated(i, allocator);
(*in_names)[i] = name.get();
}
// Output info
num_nodes = session->GetOutputCount();
out_names->resize(num_nodes);
for (int i = 0; i < num_nodes; ++i) {
std::unique_ptr<char, Ort::detail::AllocatedFree> name = session->GetOutputNameAllocated(i, allocator);
(*out_names)[i] = name.get();
}
}
void FsmnVad::Forward(
const std::vector<std::vector<float>> &chunk_feats,
std::vector<std::vector<float>> *out_prob,

5
runtime/onnxruntime/src/fsmn-vad.h
View File

@ -34,6 +34,7 @@ public:
std::shared_ptr<Ort::Session> vad_session_ = nullptr;
Ort::Env env_;
Ort::SessionOptions session_options_;
vector<string> m_strInputNames, m_strOutputNames;
std::vector<const char *> vad_in_names_;
std::vector<const char *> vad_out_names_;
std::vector<std::vector<float>> in_cache_;
@ -54,10 +55,6 @@ private:
void ReadModel(const char* vad_model);
void LoadConfigFromYaml(const char* filename);
static void GetInputOutputInfo(
const std::shared_ptr<Ort::Session> &session,
std::vector<const char *> *in_names, std::vector<const char *> *out_names);
void FbankKaldi(float sample_rate, std::vector<std::vector<float>> &vad_feats,
std::vector<float> &waves);

62
runtime/onnxruntime/src/paraformer.cpp
View File

@ -45,26 +45,8 @@ void Paraformer::InitAsr(const std::string &am_model, const std::string &am_cmvn
exit(-1);
}
string strName;
GetInputName(m_session_.get(), strName);
m_strInputNames.push_back(strName.c_str());
GetInputName(m_session_.get(), strName,1);
m_strInputNames.push_back(strName);
if (use_hotword) {
GetInputName(m_session_.get(), strName, 2);
m_strInputNames.push_back(strName);
}
size_t numOutputNodes = m_session_->GetOutputCount();
for(int index=0; index<numOutputNodes; index++){
GetOutputName(m_session_.get(), strName, index);
m_strOutputNames.push_back(strName);
}
for (auto& item : m_strInputNames)
m_szInputNames.push_back(item.c_str());
for (auto& item : m_strOutputNames)
m_szOutputNames.push_back(item.c_str());
GetInputNames(m_session_.get(), m_strInputNames, m_szInputNames);
GetOutputNames(m_session_.get(), m_strOutputNames, m_szOutputNames);
vocab = new Vocab(token_file.c_str());
phone_set_ = new PhoneSet(token_file.c_str());
LoadCmvn(am_cmvn.c_str());
@ -162,28 +144,30 @@ void Paraformer::InitAsr(const std::string &am_model, const std::string &en_mode
exit(-1);
}
string strName;
GetInputName(m_session_.get(), strName);
m_strInputNames.push_back(strName.c_str());
GetInputName(m_session_.get(), strName,1);
m_strInputNames.push_back(strName);
// string strName;
// GetInputName(m_session_.get(), strName);
// m_strInputNames.push_back(strName.c_str());
// GetInputName(m_session_.get(), strName,1);
// m_strInputNames.push_back(strName);
if (use_hotword) {
GetInputName(m_session_.get(), strName, 2);
m_strInputNames.push_back(strName);
}
// if (use_hotword) {
// GetInputName(m_session_.get(), strName, 2);
// m_strInputNames.push_back(strName);
// }
// support time stamp
size_t numOutputNodes = m_session_->GetOutputCount();
for(int index=0; index<numOutputNodes; index++){
GetOutputName(m_session_.get(), strName, index);
m_strOutputNames.push_back(strName);
}
// // support time stamp
// size_t numOutputNodes = m_session_->GetOutputCount();
// for(int index=0; index<numOutputNodes; index++){
// GetOutputName(m_session_.get(), strName, index);
// m_strOutputNames.push_back(strName);
// }
for (auto& item : m_strInputNames)
m_szInputNames.push_back(item.c_str());
for (auto& item : m_strOutputNames)
m_szOutputNames.push_back(item.c_str());
// for (auto& item : m_strInputNames)
// m_szInputNames.push_back(item.c_str());
// for (auto& item : m_strOutputNames)
// m_szOutputNames.push_back(item.c_str());
GetInputNames(m_session_.get(), m_strInputNames, m_szInputNames);
GetOutputNames(m_session_.get(), m_strOutputNames, m_szOutputNames);
}
void Paraformer::InitLm(const std::string &lm_file,

22
runtime/onnxruntime/src/sensevoice-small.cpp
View File

@ -42,26 +42,8 @@ void SenseVoiceSmall::InitAsr(const std::string &am_model, const std::string &am
exit(-1);
}
string strName;
GetInputName(m_session_.get(), strName);
m_strInputNames.push_back(strName.c_str());
GetInputName(m_session_.get(), strName,1);
m_strInputNames.push_back(strName);
GetInputName(m_session_.get(), strName,2);
m_strInputNames.push_back(strName);
GetInputName(m_session_.get(), strName,3);
m_strInputNames.push_back(strName);
size_t numOutputNodes = m_session_->GetOutputCount();
for(int index=0; index<numOutputNodes; index++){
GetOutputName(m_session_.get(), strName, index);
m_strOutputNames.push_back(strName);
}
for (auto& item : m_strInputNames)
m_szInputNames.push_back(item.c_str());
for (auto& item : m_strOutputNames)
m_szOutputNames.push_back(item.c_str());
GetInputNames(m_session_.get(), m_strInputNames, m_szInputNames);
GetOutputNames(m_session_.get(), m_strOutputNames, m_szOutputNames);
vocab = new Vocab(token_file.c_str());
LoadCmvn(am_cmvn.c_str());
}