summaryrefslogtreecommitdiff
path: root/source/use_case/kws_asr/src
diff options
context:
space:
mode:
authorKshitij Sisodia <kshitij.sisodia@arm.com>2021-12-24 11:05:11 +0000
committerLiam Barry <liam.barry@arm.com>2021-12-24 14:20:36 +0000
commit76a1580861210e0310db23acbc29e1064ae30ead (patch)
treef947145cffd944aa3724c90745fc0e9d8e2fb2f4 /source/use_case/kws_asr/src
parent871fcdc755173b9f7ecb8cf9dc8dc6306329958c (diff)
downloadml-embedded-evaluation-kit-76a1580861210e0310db23acbc29e1064ae30ead.tar.gz
MLECO-2599: Replace DSCNN with MicroNet for KWS
Added SoftMax function to Mathutils to allow MicroNet to output probability as it does not nativelu have this layer. Minor refactoring to accommodate Softmax Calculations Extensive renaming and updating of documentation and resource download script. Added SoftMax function to Mathutils to allow MicroNet to output probability. Change-Id: I7cbbda1024d14b85c9ac1beea7ca8fbffd0b6eb5 Signed-off-by: Liam Barry <liam.barry@arm.com>
Diffstat (limited to 'source/use_case/kws_asr/src')
-rw-r--r--source/use_case/kws_asr/src/AsrClassifier.cc3
-rw-r--r--source/use_case/kws_asr/src/MainLoop.cc10
-rw-r--r--source/use_case/kws_asr/src/MicroNetKwsModel.cc (renamed from source/use_case/kws_asr/src/DsCnnModel.cc)13
-rw-r--r--source/use_case/kws_asr/src/UseCaseHandler.cc20
4 files changed, 22 insertions, 24 deletions
diff --git a/source/use_case/kws_asr/src/AsrClassifier.cc b/source/use_case/kws_asr/src/AsrClassifier.cc
index 57d5058..3f9cd7b 100644
--- a/source/use_case/kws_asr/src/AsrClassifier.cc
+++ b/source/use_case/kws_asr/src/AsrClassifier.cc
@@ -73,8 +73,9 @@ template bool arm::app::AsrClassifier::GetTopResults<int8_t>(TfLiteTensor* tenso
bool arm::app::AsrClassifier::GetClassificationResults(
TfLiteTensor* outputTensor,
std::vector<ClassificationResult>& vecResults,
- const std::vector <std::string>& labels, uint32_t topNCount)
+ const std::vector <std::string>& labels, uint32_t topNCount, bool use_softmax)
{
+ UNUSED(use_softmax);
vecResults.clear();
constexpr int minTensorDims = static_cast<int>(
diff --git a/source/use_case/kws_asr/src/MainLoop.cc b/source/use_case/kws_asr/src/MainLoop.cc
index d5a2c2b..30cb084 100644
--- a/source/use_case/kws_asr/src/MainLoop.cc
+++ b/source/use_case/kws_asr/src/MainLoop.cc
@@ -16,11 +16,11 @@
*/
#include "hal.h" /* Brings in platform definitions. */
#include "InputFiles.hpp" /* For input images. */
-#include "Labels_dscnn.hpp" /* For DS-CNN label strings. */
+#include "Labels_micronetkws.hpp" /* For MicroNetKws label strings. */
#include "Labels_wav2letter.hpp" /* For Wav2Letter label strings. */
#include "Classifier.hpp" /* KWS classifier. */
#include "AsrClassifier.hpp" /* ASR classifier. */
-#include "DsCnnModel.hpp" /* KWS model class for running inference. */
+#include "MicroNetKwsModel.hpp" /* KWS model class for running inference. */
#include "Wav2LetterModel.hpp" /* ASR model class for running inference. */
#include "UseCaseCommonUtils.hpp" /* Utils functions. */
#include "UseCaseHandler.hpp" /* Handlers for different user options. */
@@ -69,7 +69,7 @@ static uint32_t GetOutputInnerLen(const arm::app::Model& model,
void main_loop(hal_platform& platform)
{
/* Model wrapper objects. */
- arm::app::DsCnnModel kwsModel;
+ arm::app::MicroNetKwsModel kwsModel;
arm::app::Wav2LetterModel asrModel;
/* Load the models. */
@@ -81,7 +81,7 @@ void main_loop(hal_platform& platform)
/* Initialise the asr model using the same allocator from KWS
* to re-use the tensor arena. */
if (!asrModel.Init(kwsModel.GetAllocator())) {
- printf_err("Failed to initalise ASR model\n");
+ printf_err("Failed to initialise ASR model\n");
return;
}
@@ -137,7 +137,7 @@ void main_loop(hal_platform& platform)
caseContext.Set<const std::vector <std::string>&>("kwslabels", kwsLabels);
/* Index of the kws outputs we trigger ASR on. */
- caseContext.Set<uint32_t>("keywordindex", 2);
+ caseContext.Set<uint32_t>("keywordindex", 9 );
/* Loop. */
bool executionSuccessful = true;
diff --git a/source/use_case/kws_asr/src/DsCnnModel.cc b/source/use_case/kws_asr/src/MicroNetKwsModel.cc
index 71d4ceb..4b44580 100644
--- a/source/use_case/kws_asr/src/DsCnnModel.cc
+++ b/source/use_case/kws_asr/src/MicroNetKwsModel.cc
@@ -14,7 +14,7 @@
* See the License for the specific language governing permissions and
* limitations under the License.
*/
-#include "DsCnnModel.hpp"
+#include "MicroNetKwsModel.hpp"
#include "hal.h"
@@ -27,21 +27,18 @@ namespace kws {
} /* namespace app */
} /* namespace arm */
-const tflite::MicroOpResolver& arm::app::DsCnnModel::GetOpResolver()
+const tflite::MicroOpResolver& arm::app::MicroNetKwsModel::GetOpResolver()
{
return this->m_opResolver;
}
-bool arm::app::DsCnnModel::EnlistOperations()
+bool arm::app::MicroNetKwsModel::EnlistOperations()
{
this->m_opResolver.AddAveragePool2D();
this->m_opResolver.AddConv2D();
this->m_opResolver.AddDepthwiseConv2D();
this->m_opResolver.AddFullyConnected();
this->m_opResolver.AddRelu();
- this->m_opResolver.AddSoftmax();
- this->m_opResolver.AddQuantize();
- this->m_opResolver.AddDequantize();
this->m_opResolver.AddReshape();
#if defined(ARM_NPU)
@@ -56,12 +53,12 @@ bool arm::app::DsCnnModel::EnlistOperations()
return true;
}
-const uint8_t* arm::app::DsCnnModel::ModelPointer()
+const uint8_t* arm::app::MicroNetKwsModel::ModelPointer()
{
return arm::app::kws::GetModelPointer();
}
-size_t arm::app::DsCnnModel::ModelSize()
+size_t arm::app::MicroNetKwsModel::ModelSize()
{
return arm::app::kws::GetModelLen();
} \ No newline at end of file
diff --git a/source/use_case/kws_asr/src/UseCaseHandler.cc b/source/use_case/kws_asr/src/UseCaseHandler.cc
index 1d88ba1..c67be22 100644
--- a/source/use_case/kws_asr/src/UseCaseHandler.cc
+++ b/source/use_case/kws_asr/src/UseCaseHandler.cc
@@ -20,8 +20,8 @@
#include "InputFiles.hpp"
#include "AudioUtils.hpp"
#include "UseCaseCommonUtils.hpp"
-#include "DsCnnModel.hpp"
-#include "DsCnnMfcc.hpp"
+#include "MicroNetKwsModel.hpp"
+#include "MicroNetKwsMfcc.hpp"
#include "Classifier.hpp"
#include "KwsResult.hpp"
#include "Wav2LetterMfcc.hpp"
@@ -77,12 +77,12 @@ namespace app {
*
* @param[in] mfcc MFCC feature calculator.
* @param[in,out] inputTensor Input tensor pointer to store calculated features.
- * @param[in] cacheSize Size of the feture vectors cache (number of feature vectors).
+ * @param[in] cacheSize Size of the feature vectors cache (number of feature vectors).
*
* @return function function to be called providing audio sample and sliding window index.
**/
static std::function<void (std::vector<int16_t>&, int, bool, size_t)>
- GetFeatureCalculator(audio::DsCnnMFCC& mfcc,
+ GetFeatureCalculator(audio::MicroNetMFCC& mfcc,
TfLiteTensor* inputTensor,
size_t cacheSize);
@@ -98,8 +98,8 @@ namespace app {
constexpr uint32_t dataPsnTxtInfStartY = 40;
constexpr int minTensorDims = static_cast<int>(
- (arm::app::DsCnnModel::ms_inputRowsIdx > arm::app::DsCnnModel::ms_inputColsIdx)?
- arm::app::DsCnnModel::ms_inputRowsIdx : arm::app::DsCnnModel::ms_inputColsIdx);
+ (arm::app::MicroNetKwsModel::ms_inputRowsIdx > arm::app::MicroNetKwsModel::ms_inputColsIdx)?
+ arm::app::MicroNetKwsModel::ms_inputRowsIdx : arm::app::MicroNetKwsModel::ms_inputColsIdx);
KWSOutput output;
@@ -128,7 +128,7 @@ namespace app {
const uint32_t kwsNumMfccFeats = ctx.Get<uint32_t>("kwsNumMfcc");
const uint32_t kwsNumAudioWindows = ctx.Get<uint32_t>("kwsNumAudioWins");
- audio::DsCnnMFCC kwsMfcc = audio::DsCnnMFCC(kwsNumMfccFeats, kwsFrameLength);
+ audio::MicroNetMFCC kwsMfcc = audio::MicroNetMFCC(kwsNumMfccFeats, kwsFrameLength);
kwsMfcc.Init();
/* Deduce the data length required for 1 KWS inference from the network parameters. */
@@ -152,7 +152,7 @@ namespace app {
/* We expect to be sampling 1 second worth of data at a time
* NOTE: This is only used for time stamp calculation. */
- const float kwsAudioParamsSecondsPerSample = 1.0/audio::DsCnnMFCC::ms_defaultSamplingFreq;
+ const float kwsAudioParamsSecondsPerSample = 1.0/audio::MicroNetMFCC::ms_defaultSamplingFreq;
auto currentIndex = ctx.Get<uint32_t>("clipIndex");
@@ -230,7 +230,7 @@ namespace app {
kwsClassifier.GetClassificationResults(
kwsOutputTensor, kwsClassificationResult,
- ctx.Get<std::vector<std::string>&>("kwslabels"), 1);
+ ctx.Get<std::vector<std::string>&>("kwslabels"), 1, true);
kwsResults.emplace_back(
kws::KwsResult(
@@ -604,7 +604,7 @@ namespace app {
static std::function<void (std::vector<int16_t>&, int, bool, size_t)>
- GetFeatureCalculator(audio::DsCnnMFCC& mfcc, TfLiteTensor* inputTensor, size_t cacheSize)
+ GetFeatureCalculator(audio::MicroNetMFCC& mfcc, TfLiteTensor* inputTensor, size_t cacheSize)
{
std::function<void (std::vector<int16_t>&, size_t, bool, size_t)> mfccFeatureCalc;