[IE Samples] New command line parameters format for speech sample (#11051)

* New command line parameters format for speech sample

* fixed notes

* changed format for scale factor

* changed format for scale factor in tests

* added more variants, when name is directy specified for i/o/r like it is done for sf

* removed nthreads flag

* fixed notes

* changed output params

* updated tests with new format

Co-authored-by: Alexander Zhogov <alexander.zhogov@intel.com>

samples/cpp/speech_sample/main.cpp

@@ -49,7 +49,10 @@ int main(int argc, char* argv[]) {
         BaseFile* fileOutput;
         ArkFile arkFile;
         NumpyFile numpyFile;
-        auto extInputFile = fileExt(FLAGS_i);
+        std::pair<std::string, std::vector<std::string>> input_data;
+        if (!FLAGS_i.empty())
+            input_data = parse_parameters(FLAGS_i);
+        auto extInputFile = fileExt(input_data.first);
         if (extInputFile == "ark") {
             file = &arkFile;
         } else if (extInputFile == "npz") {
@@ -60,9 +63,9 @@ int main(int argc, char* argv[]) {
         std::vector<std::string> inputFiles;
         std::vector<uint32_t> numBytesThisUtterance;
         uint32_t numUtterances(0);
-        if (!FLAGS_i.empty()) {
+        if (!input_data.first.empty()) {
             std::string outStr;
-            std::istringstream stream(FLAGS_i);
+            std::istringstream stream(input_data.first);
             uint32_t currentNumUtterances(0), currentNumBytesThisUtterance(0);
             while (getline(stream, outStr, ',')) {
                 std::string filename(fileNameNoExt(outStr) + "." + extInputFile);
@@ -89,8 +92,16 @@ int main(int argc, char* argv[]) {
         std::vector<std::string> output_names;
         std::vector<size_t> ports;
         // --------------------------- Processing custom outputs ---------------------------------------------
-        if (!FLAGS_oname.empty()) {
+        std::pair<std::string, std::vector<std::string>> output_data;
-            output_names = convert_str_to_vector(FLAGS_oname);
+        std::pair<std::string, std::vector<std::string>> reference_data;
+        if (!FLAGS_o.empty())
+            output_data = parse_parameters(FLAGS_o);
+        if (!FLAGS_r.empty())
+            reference_data = parse_parameters(FLAGS_r);
+        if (!output_data.second.empty())
+            output_names = output_data.second;
+        else if (!reference_data.second.empty())
+            output_names = reference_data.second;
         for (const auto& output_name : output_names) {
             auto pos_layer = output_name.rfind(":");
             if (pos_layer == std::string::npos) {
@@ -103,7 +114,6 @@ int main(int argc, char* argv[]) {
                 throw std::logic_error("Ports should have integer type");
             }
         }
-        }
         // ------------------------------ Preprocessing ------------------------------------------------------
         // the preprocessing steps can be done only for loaded network and are not applicable for the imported network
         // (already compiled)
@@ -304,8 +314,8 @@ int main(int argc, char* argv[]) {
         std::vector<ov::Tensor> ptrInputBlobs;
         auto cInputInfo = executableNet.inputs();
         check_number_of_inputs(cInputInfo.size(), numInputFiles);
-        if (!FLAGS_iname.empty()) {
+        if (!input_data.second.empty()) {
-            std::vector<std::string> inputNameBlobs = convert_str_to_vector(FLAGS_iname);
+            std::vector<std::string> inputNameBlobs = input_data.second;
             if (inputNameBlobs.size() != cInputInfo.size()) {
                 std::string errMessage(std::string("Number of network inputs ( ") + std::to_string(cInputInfo.size()) +
                                        " ) is not equal to the number of inputs entered in the -iname argument ( " +
@@ -328,15 +338,15 @@ int main(int argc, char* argv[]) {
         std::vector<std::string> output_name_files;
         std::vector<std::string> reference_name_files;
         size_t count_file = 1;
-        if (!FLAGS_o.empty()) {
+        if (!output_data.first.empty()) {
-            output_name_files = convert_str_to_vector(FLAGS_o);
+            output_name_files = convert_str_to_vector(output_data.first);
             if (output_name_files.size() != outputs.size() && !outputs.empty()) {
                 throw std::logic_error("The number of output files is not equal to the number of network outputs.");
             }
             count_file = output_name_files.empty() ? 1 : output_name_files.size();
         }
-        if (!FLAGS_r.empty()) {
+        if (!reference_data.first.empty()) {
-            reference_name_files = convert_str_to_vector(FLAGS_r);
+            reference_name_files = convert_str_to_vector(reference_data.first);
             if (reference_name_files.size() != outputs.size() && !outputs.empty()) {
                 throw std::logic_error("The number of reference files is not equal to the number of network outputs.");
             }
@@ -429,9 +439,9 @@ int main(int argc, char* argv[]) {
             BaseFile* fileReferenceScores;
             std::string refUtteranceName;
 
-            if (!FLAGS_r.empty()) {
+            if (!reference_data.first.empty()) {
                 /** Read file with reference scores **/
-                auto exReferenceScoresFile = fileExt(FLAGS_r);
+                auto exReferenceScoresFile = fileExt(reference_data.first);
                 if (exReferenceScoresFile == "ark") {
                     fileReferenceScores = &arkFile;
                 } else if (exReferenceScoresFile == "npz") {
@@ -540,12 +550,12 @@ int main(int argc, char* argv[]) {
                         continue;
                     }
                     ptrInputBlobs.clear();
-                    if (FLAGS_iname.empty()) {
+                    if (input_data.second.empty()) {
                         for (auto& input : cInputInfo) {
                             ptrInputBlobs.push_back(inferRequest.inferRequest.get_tensor(input));
                         }
                     } else {
-                        std::vector<std::string> inputNameBlobs = convert_str_to_vector(FLAGS_iname);
+                        std::vector<std::string> inputNameBlobs = input_data.second;
                         for (const auto& input : inputNameBlobs) {
                             ov::Tensor blob = inferRequests.begin()->inferRequest.get_tensor(input);
                             if (!blob) {
@@ -638,7 +648,7 @@ int main(int argc, char* argv[]) {
 
             for (size_t next_output = 0; next_output < count_file; next_output++) {
                 if (!FLAGS_o.empty()) {
-                    auto exOutputScoresFile = fileExt(FLAGS_o);
+                    auto exOutputScoresFile = fileExt(output_data.first);
                     if (exOutputScoresFile == "ark") {
                         fileOutput = &arkFile;
                     } else if (exOutputScoresFile == "npz") {

samples/cpp/speech_sample/speech_sample.hpp

@@ -96,10 +96,6 @@ static const char scale_factor_message[] =
 /// @brief message for batch size argument
 static const char batch_size_message[] = "Optional. Batch size 1-8 (default 1)";
 
-/// @brief message for #threads for CPU inference
-static const char infer_num_threads_message[] = "Optional. Number of threads to use for concurrent async"
-                                                " inference requests on the GNA.";
-
 /// @brief message for left context window argument
 static const char context_window_message_l[] =
     "Optional. Number of frames for left context windows (default is 0). "
@@ -184,9 +180,6 @@ DEFINE_string(sf, "", scale_factor_message);
 /// @brief Batch size (default 0)
 DEFINE_int32(bs, 0, batch_size_message);
 
-/// @brief Number of threads to use for inference on the CPU (also affects Hetero cases)
-DEFINE_int32(nthreads, 1, infer_num_threads_message);
-
 /// @brief Right context window size (default 0)
 DEFINE_int32(cw_r, 0, context_window_message_r);
 

samples/cpp/speech_sample/utils.hpp

@@ -432,7 +432,7 @@ bool check_name(const ov::OutputVector& nodes, const std::string& node_name) {
 
 /**
  * @brief Parse scale factors per input
- * Format : <input_name1>:<sf1>,<input2>:<sf2> or just <sf>
+ * Format : <input_name1>=<sf1>,<input2>=<sf2> or just <sf>
  * @param inputs model inputs
  * @param values_string values_string input string
  * @return map of scale factors per input
@@ -454,11 +454,11 @@ std::map<std::string, float> parse_scale_factors(const ov::OutputVector& inputs,
     std::map<std::string, float> result;
     auto scale_factor_strings = split(values_string, ',');
     for (auto& scale_factor_string : scale_factor_strings) {
-        auto values = split(scale_factor_string, ':');
+        auto values = split(scale_factor_string, '=');
         if (values.size() == 1) {
             if (scale_factor_strings.size() != 1) {
                 throw std::logic_error("Unrecognized scale factor format! "
-                                       "Please specify <input_name1>:<sf1>,<input_name2>:<sf2> or "
+                                       "Please specify <input_name1>=<sf1>,<input_name2>=<sf2> or "
                                        "just <sf> to be applied to all inputs");
             }
             auto scale_factor = get_sf(values.at(0));
@@ -468,8 +468,7 @@ std::map<std::string, float> parse_scale_factors(const ov::OutputVector& inputs,
         } else if (values.size() > 0) {
             auto sf_sting = values.back();
             values.pop_back();
-            // input name can contain port, concat back
+            auto input_name = values.back();
-            auto input_name = concat(values, ':');
             check_name(inputs, input_name);
             result[input_name] = get_sf(sf_sting, input_name);
         }
@@ -535,3 +534,37 @@ std::map<std::string, std::string> parse_input_layouts(const std::string& layout
         throw std::logic_error("Can't parse input parameter string: " + layout_string);
     return return_value;
 }
+
+/**
+ * @brief Parse parameters for inputs/outputs like as "<name1>=<file1.ark/.npz>,<name2>=<file2.ark/.npz>" or
+ * "<file.ark/.npz>" in case of one input/output
+ * @param file_paths_string input/output path
+ * @return pair of filename and vector of tensor_names
+ */
+std::pair<std::string, std::vector<std::string>> parse_parameters(const std::string file_paths_string) {
+    auto search_string = file_paths_string;
+    char comma_delim = ',';
+    char equal_delim = '=';
+    std::string filename = "";
+    std::vector<std::string> tensor_names;
+    std::vector<std::string> filenames;
+    if (!std::count(search_string.begin(), search_string.end(), comma_delim) &&
+        !std::count(search_string.begin(), search_string.end(), equal_delim)) {
+        return {search_string, tensor_names};
+    }
+    search_string += comma_delim;
+    std::vector<std::string> splitted = split(search_string, comma_delim);
+    for (size_t j = 0; j < splitted.size(); j++) {
+        auto semicolon_pos = splitted[j].find_first_of(equal_delim);
+        if (semicolon_pos != std::string::npos) {
+            tensor_names.push_back(splitted[j].substr(0, semicolon_pos));
+            filenames.push_back(splitted[j].substr(semicolon_pos + 1, std::string::npos));
+        }
+    }
+    for (std::vector<std::string>::const_iterator name = filenames.begin(); name != filenames.end(); ++name) {
+        filename += *name;
+        if (name != filenames.end() - 1)
+            filename += comma_delim;
+    }
+    return {filename, tensor_names};
+}

tests/samples_tests/smoke_tests/common/samples_common_test_clas.py

@@ -186,16 +186,16 @@ class SamplesCommonTestClass():
         return model
 
     @staticmethod
-    def join_env_path(param, executable_path):
+    def join_env_path(param, executable_path, complete_path=True):
         gpu_lib_path = os.path.join(os.environ.get('IE_APP_PATH'), 'lib')
         if 'i' in param:
             # If batch > 1, then concatenate images
             if ' ' in param['i']:
                 param['i'] = param['i'].split(' ')
-            else:
+            elif complete_path:
                 param['i'] = list([param['i']])
         for k in param.keys():
-            if ('i' == k):
+            if ('i' == k) and complete_path:
                 param['i'] = [os.path.join(Environment.env['test_data'], e) for e in param['i']]
                 param['i'] = ' '.join(param['i'])
             elif ('ref_m' == k):
@@ -236,10 +236,10 @@ class SamplesCommonTestClass():
                 param['l'] = os.path.join(Environment.env['test_data'], param['l'])
             elif ('pp' == k):
                 param['pp'] = gpu_lib_path
-            elif ('r' == k):
+            elif ('r' == k) and complete_path:
                 if len(param['r']) > 0:
                     param['r'] = os.path.join(Environment.env['test_data'], param['r'])
-            elif ('o' == k):
+            elif ('o' == k) and complete_path:
                 param['o'] = os.path.join(Environment.env['out_directory'], param['o'])
             elif ('wg' == k):
                 param['wg'] = os.path.join(Environment.env['out_directory'], param['wg'])
@@ -344,7 +344,7 @@ class SamplesCommonTestClass():
             "Path for test data {} is not exist!".format(Environment.env['test_data'])
         cls.output_dir = Environment.env['out_directory']
 
-    def _test(self, param, use_preffix=True, get_cmd_func=None, get_shell_result=False, long_hyphen=None):
+    def _test(self, param, use_preffix=True, get_cmd_func=None, get_shell_result=False, long_hyphen=None, complete_path=True):
         """
         :param param:
         :param use_preffix: use it when sample doesn't require keys (i.e. hello_classification <path_to_model> <path_to_image>
@@ -379,7 +379,7 @@ class SamplesCommonTestClass():
         if get_cmd_func is None:
             get_cmd_func = self.get_cmd_line
 
-        self.join_env_path(param_cp, executable_path=self.executable_path)
+        self.join_env_path(param_cp, executable_path=self.executable_path, complete_path=complete_path)
 
         # Updating all attributes in the original dictionary (param), because param_cp was changes (join_env_path)
         for key in param.keys():

tests/samples_tests/smoke_tests/test_speech_sample.py

@@ -15,6 +15,7 @@ import pytest
 import sys
 import logging as log
 from common.samples_common_test_clas import SamplesCommonTestClass
+from common.samples_common_test_clas import Environment
 from common.samples_common_test_clas import get_tests
 from common.common_utils import parse_avg_err
 
@@ -27,12 +28,25 @@ test_data = get_tests(cmd_params={'i': [os.path.join('ark', 'dev93_10.ark')],
                                            'o': ['res_output.ark'],
                                            'r': [os.path.join('ark', 'dev93_scores_10.ark')],
                                            'qb': [8, 16],
-                                           'sf': ["Parameter:2175.43", "2175.43"],
+                                           'sf': ["2175.43"],
                                            'q': ["static", "user"],
                                            'd': ['GNA_SW_EXACT']},
                                use_device=False
                                )
 							   
+new_format_test_data = get_tests(cmd_params={'i': ['Parameter=' + os.path.join(Environment.env['test_data'], 'ark', 'dev93_10.ark')],
+                                           'm': [os.path.join('wsj', 'FP32', 'wsj_dnn5b.xml')],
+                                           'layout': ["[NC]"],
+                                           'bs': [1],
+                                           'o': ['affinetransform14/Fused_Add_:0=' + os.path.join(Environment.env['test_data'], 'res_output.ark')],
+                                           'r': ['affinetransform14/Fused_Add_:0=' + os.path.join(Environment.env['test_data'], 'ark', 'dev93_scores_10.ark')],
+                                           'qb': [8],
+                                           'sf': ["Parameter=2175.43"],
+                                           'q': ["static"],
+                                           'd': ['GNA_SW_EXACT']},
+                               use_device=False
+                               )
+
 class TestSpeechSample(SamplesCommonTestClass):
     @classmethod
     def setup_class(cls):
@@ -43,7 +57,14 @@ class TestSpeechSample(SamplesCommonTestClass):
     @pytest.mark.parametrize("param", test_data)
     def test_speech_sample_nthreads(self, param):
         stdout = self._test(param).split('\n')
-        assert os.path.isfile(param['o']), "Ark file after infer was not found"
+
+        avg_error = parse_avg_err(stdout)
+        log.info('Average scores diff: {}'.format(avg_error))
+        assert avg_error <= self.threshold
+		
+    @pytest.mark.parametrize("param", new_format_test_data)
+    def test_speech_sample_new_format(self, param):
+        stdout = self._test(param, complete_path=False).split('\n')
 
         avg_error = parse_avg_err(stdout)
         log.info('Average scores diff: {}'.format(avg_error))