This document describes the process of setting up and running the Arm® Ethos™-U55 Keyword Spotting example.
Use case code could be found in source/use_case/kws directory.
The DS-CNN keyword spotting model that is supplied with the Code Samples expects audio data to be preprocessed in a specific way before performing an inference. This section aims to provide an overview of the feature extraction process used.
First the audio data is normalized to the range (-1, 1).
Note: Mel-frequency cepstral coefficients (MFCCs) are a common feature extracted from audio data and can be used as input for machine learning tasks like keyword spotting and speech recognition. See source/application/main/include/Mfcc.hpp for implementation details.
Next, a window of 640 audio samples is taken from the start of the audio clip. From these 640 samples we calculate 10 MFCC features.
The whole window is shifted to the right by 320 audio samples and 10 new MFCC features are calculated. This process of shifting and calculating is repeated until the end of the 16000 audio samples needed to perform an inference is reached. In total this will be 49 windows that each have 10 MFCC features calculated for them, giving an input tensor of shape 49x10.
These extracted features are quantized, and an inference is performed.
If the audio clip is longer than 16000 audio samples then the initial starting position is offset by 16000/2 = 8000 audio samples. From this new starting point, MFCC features for the next 16000 audio samples are calculated and another inference is performed (i.e. do an inference for samples 8000-24000).
Note: Parameters of the MFCC feature extraction such as window size, stride, number of features etc. all depend on what was used during model training. These values are specific to each model and if you try a different keyword spotting model that uses MFCC input then values are likely to need changing to match the new model. In addition, MFCC feature extraction methods can vary slightly with different normalization methods or scaling etc. being used.
After an inference is complete the highest probability detected word is output to console, providing its probability is larger than a threshold value (default 0.9).
If multiple inferences are performed for an audio clip, then multiple results will be output.
In addition to the already specified build option in the main documentation, keyword spotting use case adds:
kws_MODEL_TFLITE_PATH - Path to the NN model file in TFLite format. Model will be processed and included into the application axf file. The default value points to one of the delivered set of models. Note that the parameters
ETHOS_U55_ENABLED should be aligned with the chosen model, i.e.:
ETHOS_U55_ENABLED is set to
1, the NN model is assumed to be optimized. The model will naturally fall back to the Arm® Cortex®-M CPU if an unoptimized model is supplied.
ETHOS_U55_ENABLED is set to
0, the NN model is assumed to be unoptimized. Supplying an optimized model in this case will result in a runtime error.
kws_FILE_PATH: Path to the directory containing audio files, or a path to single WAV file, to be used in the application. The default value points to the resources/kws/samples folder containing the delivered set of audio clips.
kws_LABELS_TXT_FILE: Path to the labels' text file. The file is used to map key word class index to the text label. The default value points to the delivered labels.txt file inside the delivery package.
kws_AUDIO_RATE: Input data sampling rate. Each audio file from kws_FILE_PATH is preprocessed during the build to match NN model input requirements. Default value is 16000.
kws_AUDIO_MONO: If set to ON the audio data will be converted to mono. Default is ON.
kws_AUDIO_OFFSET: Start loading audio data starting from this offset (in seconds). Default value is 0.
kws_AUDIO_DURATION: Length of the audio data to be used in the application in seconds. Default is 0 meaning the whole audio file will be taken.
kws_AUDIO_MIN_SAMPLES: Minimum number of samples required by the network model. If the audio clip is shorter than this number, it is padded with zeros. Default value is 16000.
kws_MODEL_SCORE_THRESHOLD: Threshold value [0.0, 1.0] that must be applied to the inference results for a label to be deemed valid. Default is 0.9
kws_ACTIVATION_BUF_SZ: The intermediate/activation buffer size reserved for the NN model. By default, it is set to 1MiB and should be enough for most models.
In order to build ONLY keyword spotting example application add to the
cmake command line specified in Building
Note: This section describes the process for configuring the build for
MPS3: SSE-300for different target platform see Building section.
In order to build only the keyword spotting example, create a build directory and navigate inside, for example:
mkdir build_kws && cd build_kws
On Linux, execute the following command to build Keyword Spotting application to run on the Ethos-U55 Fast Model when providing only the mandatory arguments for CMake configuration:
cmake ../ -DUSE_CASE_BUILD=kws
To configure a build that can be debugged using Arm-DS, we can just specify the build type as
Debug and use the
Arm Compiler toolchain file:
cmake .. \ -DCMAKE_TOOLCHAIN_FILE=scripts/cmake/toolchains/bare-metal-armclang.cmake \ -DCMAKE_BUILD_TYPE=Debug \ -DUSE_CASE_BUILD=kws
Note: If re-building with changed parameters values, it is highly advised to clean the build directory and re-run the CMake command.
If the CMake command succeeded, build the application as follows:
Add VERBOSE=1 to see compilation and link details.
Results of the build will be placed under
bin ├── ethos-u-kws.axf ├── ethos-u-kws.htm ├── ethos-u-kws.map └── sectors ├── images.txt └── kws ├── dram.bin └── itcm.bin
ethos-u-kws.axf: The built application binary for the Keyword Spotting use case.
ethos-u-kws.map: Information from building the application (e.g. libraries used, what was optimized, location of objects)
ethos-u-kws.htm: Human readable file containing the call graph of application functions.
sectors/kws: Folder containing the built application, split into files for loading into different FPGA memory regions.
sectors/images.txt: Tells the FPGA which memory regions to use for loading the binaries in sectors/** folder.
The application performs inference on audio data found in the folder, or an individual file, set by the CMake parameter
To run the application with your own audio clips first create a folder to hold them and then copy the custom audio files into this folder, for example:
mkdir /tmp/custom_wavs cp my_clip.wav /tmp/custom_wavs/
Note: Clean the build directory before re-running the CMake command.
kws_FILE_PATH to the location of this folder when building:
cmake .. \ -Dkws_FILE_PATH=/tmp/custom_wavs/ \ -DUSE_CASE_BUILD=kws
The audio clips found in the
kws_FILE_PATH folder will be picked up and automatically converted to C++ files during the CMake configuration stage and then compiled into the application during the build phase for performing inference with.
The log from the configuration stage should tell you what audio clip directory path has been used:
-- User option kws_FILE_PATH is set to /tmp/custom_wavs -- Generating audio files from /tmp/custom_wavs ++ Converting my_clip.wav to my_clip.cc ++ Generating build/generated/kws/include/AudioClips.hpp ++ Generating build/generated/kws/src/AudioClips.cc -- Defined build user options: -- kws_FILE_PATH=/tmp/custom_wavs
After compiling, your custom inputs will have now replaced the default ones in the application.
Note: The CMake parameter
kws_AUDIO_MIN_SAMPLESdetermine the minimum number of input sample. When building the application, if the size of the audio clips is less then
kws_AUDIO_MIN_SAMPLESthen it will be padded so that it does.
The application performs inference using the model pointed to by the CMake parameter
Note: If you want to run the model using Ethos-U55, ensure your custom model has been run through the Vela compiler successfully before continuing. See Optimize model with Vela compiler.
To run the application with a custom model you will need to provide a labels_<model_name>.txt file of labels associated with the model. Each line of the file should correspond to one of the outputs in your model. See the provided ds_cnn_labels.txt file for an example.
Then, you must set kws_MODEL_TFLITE_PATH to the location of the Vela processed model file and kws_LABELS_TXT_FILE to the location of the associated labels file.
cmake .. \ -Dkws_MODEL_TFLITE_PATH=<path/to/custom_model_after_vela.tflite> \ -Dkws_LABELS_TXT_FILE=<path/to/labels_custom_model.txt> \ -DUSE_CASE_BUILD=kws
Note: Clean the build directory before re-running the CMake command.
.tflite model file pointed to by
kws_MODEL_TFLITE_PATH and labels text file pointed to by
kws_LABELS_TXT_FILE will be converted to C++ files during the CMake configuration stage and then compiled into the application for performing inference with.
The log from the configuration stage should tell you what model path and labels file have been used:
-- User option kws_MODEL_TFLITE_PATH is set to <path/to/custom_model_after_vela.tflite> ... -- User option kws_LABELS_TXT_FILE is set to <path/to/labels_custom_model.txt> ... -- Using <path/to/custom_model_after_vela.tflite> ++ Converting custom_model_after_vela.tflite to\ custom_model_after_vela.tflite.cc -- Generating labels file from <path/to/labels_custom_model.txt> -- writing to <path/to/build/generated/src/Labels.cc> ...
After compiling, your custom model will have now replaced the default one in the application.
The FVP is available publicly from Arm Ecosystem FVP downloads.
For Ethos-U55 evaluation, please download the MPS3 version of the Arm® Corstone™-300 model that contains Ethos-U55 and Cortex-M55. The model is currently only supported on Linux based machines. To install the FVP:
Unpack the archive
Run the install script in the extracted package
Once completed the building step, application binary ethos-u-kws.axf can be found in the
build/bin folder. Assuming the install location of the FVP was set to ~/FVP_install_location, the simulation can be started by:
A log output should appear on the terminal:
telnetterminal0: Listening for serial connection on port 5000 telnetterminal1: Listening for serial connection on port 5001 telnetterminal2: Listening for serial connection on port 5002 telnetterminal5: Listening for serial connection on port 5003
This will also launch a telnet window with the sample application's standard output and error log entries containing information about the pre-built application version, TensorFlow Lite Micro library version used, data type as well as the input and output tensor sizes of the model compiled into the executable binary.
After the application has started if
kws_FILE_PATH pointed to a single file (or a folder containing a single input file) the inference starts immediately. In case of multiple inputs choice, it outputs a menu and waits for the user input from telnet terminal:
User input required Enter option number from: 1. Classify next audio clip 2. Classify audio clip at chosen index 3. Run classification on all audio clips 4. Show NN model info 5. List audio clips Choice:
“Classify next audio clip” menu option will run inference on the next in line voice clip from the collection of the compiled audio.
Note: Note that if the clip is over a certain length, the application will invoke multiple inference runs to cover the entire file.
“Classify audio clip at chosen index” menu option will run inference on the chosen audio clip.
Note: Please make sure to select audio clip index in the range of supplied audio clips during application build. By default, pre-built application has 4 files, indexes from 0 to 3.
“Run classification on all audio clips” menu option triggers sequential inference executions on all built-in voice samples.
“Show NN model info” menu option prints information about model data type, input and output tensor sizes:
INFO - uTFL version: 2.5.0 INFO - Model info: INFO - Model INPUT tensors: INFO - tensor type is INT8 INFO - tensor occupies 490 bytes with dimensions INFO - 0: 1 INFO - 1: 1 INFO - 2: 49 INFO - 3: 10 INFO - Quant dimension: 0 INFO - Scale = 1.107164 INFO - ZeroPoint = 95 INFO - Model OUTPUT tensors: INFO - tensor type is INT8 INFO - tensor occupies 12 bytes with dimensions INFO - 0: 1 INFO - 1: 12 INFO - Quant dimension: 0 INFO - Scale = 0.003906 INFO - ZeroPoint = -128 INFO - Activation buffer (a.k.a tensor arena) size used: 72848 INFO - Number of operators: 1 INFO - Operator 0: ethos-u
“List audio clips” menu option prints a list of pair audio indexes - the original filenames embedded in the application:
[INFO] List of Files: [INFO] 0 => down.wav [INFO] 1 => right_left_up.wav [INFO] 2 => yes.wav [INFO] 3 => yes_no_go_stop.wav
Selecting the first option will run inference on the first file.
The following example illustrates application output for classification:
INFO - Inference 1/1 INFO - Final results: INFO - Total number of inferences: 1 INFO - For timestamp: 0.000000 (inference #: 0); label: down, score: 0.996094; threshold: 0.900000 INFO - Profile for Inference: INFO - NPU AXI0_RD_DATA_BEAT_RECEIVED beats: 217385 INFO - NPU AXI0_WR_DATA_BEAT_WRITTEN beats: 82607 INFO - NPU AXI1_RD_DATA_BEAT_RECEIVED beats: 59608 INFO - NPU ACTIVE cycles: 680611 INFO - NPU IDLE cycles: 561 INFO - NPU total cycles: 681172
Each inference should take less than 30 seconds on most systems running Fast Model. The profiling section of the log shows that for this inference:
Ethos-U55's PMU report:
681,172 total cycle: The number of NPU cycles
680,611 active cycles: The number of NPU cycles that were used for computation
561 idle cycles: number of cycles for which the NPU was idle
217,385 AXI0 read beats: The number of AXI beats with read transactions from AXI0 bus. AXI0 is the bus where Ethos-U55 NPU reads and writes to the computation buffers (activation buf/tensor arenas).
82,607 write cycles: The number of AXI beats with write transactions to AXI0 bus.
59,608 AXI1 read beats: The number of AXI beats with read transactions from AXI1 bus. AXI1 is the bus where Ethos-U55 NPU reads the model (read only)
For FPGA platforms, CPU cycle count can also be enabled. For FVP, however, CPU cycle counters should not be used as the CPU model is not cycle-approximate or cycle-accurate.
The application prints the highest confidence score and the associated label from ds_cnn_labels.txt file.