openvino/samples/cpp/speech_sample/README.md

# Automatic Speech Recognition C++ Sample {#openvino_inference_engine_samples_speech_sample_README}

This sample demonstrates how to execute an Asynchronous Inference of acoustic model based on Kaldi\* neural networks and speech feature vectors.

The sample works with Kaldi ARK or Numpy* uncompressed NPZ files, so it does not cover an end-to-end speech recognition scenario (speech to text), requiring additional preprocessing (feature extraction) to get a feature vector from a speech signal, as well as postprocessing (decoding) to produce text from scores.

The following C++ API is used in the application:

| Feature | API | Description |
| :--- | :--- | :--- |
| Available Devices | `ov::Core::get_available_devices`, `ov::Core::get_property` | Get information of the devices for inference |
| Import/Export Model | `ov::Core::import_model`, `ov::CompiledModel::export_model` | The GNA plugin supports loading and saving of the GNA-optimized model |
| Model Operations | `ov::set_batch`, `ov::Model::add_output`, `ov::CompiledModel::inputs`, `ov::CompiledModel::outputs` | Managing of model: configure batch_size, input and output tensors |
| Node Operations | `ov::OutputVector::size`, `ov::Output::get_shape` | Get node shape |
| Asynchronous Infer | `ov::InferRequest::start_async`, `ov::InferRequest::wait` | Do asynchronous inference and waits until inference result becomes available |
| InferRequest Operations | `ov::InferRequest::query_state`, `ov::VariableState::reset` | Gets and resets CompiledModel state control |
| Tensor Operations | `ov::Tensor::get_size`, `ov::Tensor::data`, `ov::InferRequest::get_tensor` | Get a tensor, its size and data |
| Profiling | `ov::InferRequest::get_profiling_info` | Get infer request profiling info |

Basic OpenVINO™ Runtime API is covered by [Hello Classification C++ sample](../hello_classification/README.md).

| Options | Values |
| :--- | :--- |
| Validated Models | Acoustic model based on Kaldi\* neural networks (see [Model Preparation](#model-preparation) section) |
| Model Format | OpenVINO™ toolkit Intermediate Representation (\*.xml + \*.bin) |
| Supported devices | See [Execution Modes](#execution-modes) section below and [List Supported Devices](../../../docs/OV_Runtime_UG/supported_plugins/Supported_Devices.md) |

## How It Works

At startup, the sample application reads command-line parameters, loads a specified model and input data to the OpenVINO™ Runtime plugin, performs inference on all speech utterances stored in the input file(s), logging each step in a standard output stream.  
If the `-r` option is given, error statistics are provided for each speech utterance as shown above.

You can see the explicit description of
each sample step at [Integration Steps](../../../docs/OV_Runtime_UG/Integrate_with_customer_application_new_API.md) section of "Integrate the OpenVINO™ Runtime with Your Application" guide.

### GNA-specific details

#### Quantization

If the GNA device is selected (for example, using the `-d` GNA flag), the GNA OpenVINO™ Runtime plugin quantizes the model and input feature vector sequence to integer representation before performing inference.
Several parameters control neural network quantization. The `-q` flag determines the quantization mode.
Three modes are supported:

- *static* - The first utterance in the input file is scanned for dynamic range. The scale factor (floating point scalar multiplier) required to scale the maximum input value of the first utterance to 16384 (15 bits) is used for all subsequent inputs. The neural network is quantized to accommodate the scaled input dynamic range.
- *dynamic* - The scale factor for each input batch is computed just before inference on that batch. The input and network are (re)quantized on the fly using an efficient procedure.
- *user-defined* - The user may specify a scale factor via the `-sf` flag that will be used for static quantization.

The `-qb` flag provides a hint to the GNA plugin regarding the preferred target weight resolution for all layers. For example, when `-qb 8` is specified, the plugin will use 8-bit weights wherever possible in the
network.

> **NOTE**:
>
> - It is not always possible to use 8-bit weights due to GNA hardware limitations. For example, convolutional layers always use 16-bit weights (GNA hardware version 1 and 2). This limitation will be removed in GNA hardware version 3 and higher.

#### Execution Modes

Several execution modes are supported via the `-d` flag:

- `CPU` - All calculation are performed on CPU device using CPU Plugin.
- `GPU` - All calculation are performed on GPU device using GPU Plugin.
- `MYRIAD` - All calculation are performed on Intel® Neural Compute Stick 2 device using VPU MYRIAD Plugin.
- `GNA_AUTO` - GNA hardware is used if available and the driver is installed. Otherwise, the GNA device is emulated in fast-but-not-bit-exact mode.
- `GNA_HW` - GNA hardware is used if available and the driver is installed. Otherwise, an error will occur.
- `GNA_SW` - Deprecated. The GNA device is emulated in fast-but-not-bit-exact mode.
- `GNA_SW_FP32` - Substitutes parameters and calculations from low precision to floating point (FP32).
- `GNA_SW_EXACT` - GNA device is emulated in bit-exact mode.

#### Loading and Saving Models

The GNA plugin supports loading and saving of the GNA-optimized model (non-IR) via the `-rg` and `-wg` flags.  Thereby, it is possible to avoid the cost of full model quantization at run time. The GNA plugin also supports export of firmware-compatible embedded model images for the Intel® Speech Enabling Developer Kit and Amazon Alexa* Premium Far-Field Voice Development Kit via the `-we` flag (save only).

In addition to performing inference directly from a GNA model file, these combinations of options make it possible to:

- Convert from IR format to GNA format model file (`-m`, `-wg`)
- Convert from IR format to embedded format model file (`-m`, `-we`)
- Convert from GNA format to embedded format model file (`-rg`, `-we`)

## Building

To build the sample, please use instructions available at [Build the Sample Applications](../../../docs/OV_Runtime_UG/Samples_Overview.md) section in OpenVINO™ Toolkit Samples guide.

## Running

Run the application with the -h option to see the usage message:

```
speech_sample -h
```

Usage message:

```
[ INFO ] OpenVINO Runtime version ......... <version>
[ INFO ] Build ........... <build>
[ INFO ]
[ INFO ] Parsing input parameters

speech_sample [OPTION]
Options:

    -h                         Print a usage message.
    -i "<path>"                Required. Paths to input files. Example of usage: <file1.ark,file2.ark> or <file.ark> or <file.npz>.
    -m "<path>"                Required. Path to an .xml file with a trained model (required if -rg is missing).
    -o "<path>"                Optional. Output file name to save scores. Example of usage: <output.ark> or <output.npz>
    -d "<device>"              Optional. Specify a target device to infer on. CPU, GPU, MYRIAD, GNA_AUTO, GNA_HW, GNA_HW_WITH_SW_FBACK, GNA_SW_FP32, GNA_SW_EXACT and HETERO with combination of GNA as the primary device and CPU as a secondary (e.g. HETERO:GNA,CPU) are supported. The sample will look for a suitable plugin for device specified.
    -pc                        Optional. Enables per-layer performance report.
    -q "<mode>"                Optional. Input quantization mode:  static (default), dynamic, or user (use with -sf).
    -qb "<integer>"            Optional. Weight bits for quantization: 8 or 16 (default)
    -sf "<double>"             Optional. User-specified input scale factor for quantization (use with -q user). If the network contains multiple inputs, provide scale factors by separating them with commas.
    -bs "<integer>"            Optional. Batch size 1-8
    -layout "<string>"         Optional. Prompts how network layouts should be treated by application.For example, \"input1[NCHW],input2[NC]\" or \"[NCHW]\" in case of one input size.
    -r "<path>"                Optional. Read reference score file and compare scores. Example of usage: <reference.ark> or <reference.npz>
    -rg "<path>"               Read GNA model from file using path/filename provided (required if -m is missing).
    -wg "<path>"               Optional. Write GNA model to file using path/filename provided.
    -we "<path>"               Optional. Write GNA embedded model to file using path/filename provided.
    -cw_l "<integer>"          Optional. Number of frames for left context windows (default is 0). Works only with context window networks. If you use the cw_l or cw_r flag, then batch size argument is ignored.
    -cw_r "<integer>"          Optional. Number of frames for right context windows (default is 0). Works only with context window networks. If you use the cw_r or cw_l flag, then batch size argument is ignored.
    -oname "<string>"          Optional. Layer names for output blobs. The names are separated with "," Example: Output1:port,Output2:port
    -iname "<string>"          Optional. Layer names for input blobs. The names are separated with "," Example: Input1,Input2
    -pwl_me "<double>"         Optional. The maximum percent of error for PWL function.The value must be in <0, 100> range. The default value is 1.0.
    -exec_target "<string>"    Optional. Specify GNA execution target generation. May be one of GNA_TARGET_2_0, GNA_TARGET_3_0. By default, generation corresponds to the GNA HW available in the system or the latest fully supported generation by the software. See the GNA Plugin's GNA_EXEC_TARGET config option description.
    -compile_target "<string>" Optional. Specify GNA compile target generation. May be one of GNA_TARGET_2_0, GNA_TARGET_3_0. By default, generation corresponds to the GNA HW available in the system or the latest fully supported generation by the software. See the GNA Plugin's GNA_COMPILE_TARGET config option description.

Available target devices:  CPU  GNA  GPU
```

### Model Preparation

You can use the following model optimizer command to convert a Kaldi nnet1 or nnet2 neural model to OpenVINO™ toolkit Intermediate Representation format:

```
mo --framework kaldi --input_model wsj_dnn5b.nnet --counts wsj_dnn5b.counts --remove_output_softmax --output_dir <OUTPUT_MODEL_DIR>
```

The following pre-trained models are available:

- wsj_dnn5b_smbr
- rm_lstm4f
- rm_cnn4a_smbr

All of them can be downloaded from [https://storage.openvinotoolkit.org/models_contrib/speech/2021.2](https://storage.openvinotoolkit.org/models_contrib/speech/2021.2).

### Speech Inference

Once the IR is created, you can do inference on Intel® Processors with the GNA co-processor (or emulation library):

```
speech_sample -m wsj_dnn5b.xml -i dev93_10.ark -r dev93_scores_10.ark -d GNA_AUTO -o result.ark
```

Here, the floating point Kaldi-generated reference neural network scores (`dev93_scores_10.ark`) corresponding to the input feature file (`dev93_10.ark`) are assumed to be available for comparison.

> **NOTES**:
>
> - Before running the sample with a trained model, make sure the model is converted to the intermediate representation (IR) format (\*.xml + \*.bin) using the [Model Optimizer tool](../../../docs/MO_DG/Deep_Learning_Model_Optimizer_DevGuide.md).
>
> - The sample supports input and output in numpy file format (.npz)

## Sample Output

The sample application logs each step in a standard output stream.

```
[ INFO ] OpenVINO runtime: OpenVINO Runtime version ......... 2022.1.0
[ INFO ] Build ........... 2022.1.0-6311-a90bb1ff017
[ INFO ]
[ INFO ] Parsing input parameters
[ INFO ] Loading model files:
[ INFO ] \test_data\models\wsj_dnn5b_smbr_fp32\wsj_dnn5b_smbr_fp32.xml
[ INFO ] Using scale factor of 2175.43 calculated from first utterance.
[ INFO ] Model loading time 0.0034 ms
[ INFO ] Loading model to the device GNA_AUTO
[ INFO ] Loading model to the device
[ INFO ] Number scores per frame : 3425
Utterance 0:
Total time in Infer (HW and SW):        5687.53 ms
Frames in utterance:                    1294 frames
Average Infer time per frame:           4.39531 ms
         max error: 0.705184
         avg error: 0.0448388
     avg rms error: 0.0574098
       stdev error: 0.0371649


End of Utterance 0

[ INFO ] Number scores per frame : 3425
Utterance 1:
Total time in Infer (HW and SW):        4341.34 ms
Frames in utterance:                    1005 frames
Average Infer time per frame:           4.31974 ms
         max error: 0.757597
         avg error: 0.0452166
     avg rms error: 0.0578436
       stdev error: 0.0372769


End of Utterance 1

...
End of Utterance X

[ INFO ] Execution successful
```

## Use of Sample in Kaldi* Speech Recognition Pipeline

The Wall Street Journal DNN model used in this example was prepared using the Kaldi s5 recipe and the Kaldi Nnet (nnet1) framework. It is possible to recognize speech by substituting the `speech_sample` for
Kaldi's nnet-forward command. Since the `speech_sample` does not yet use pipes, it is necessary to use temporary files for speaker-transformed feature vectors and scores when running the Kaldi speech recognition pipeline. The following operations assume that feature extraction was already performed according to the `s5` recipe and that the working directory within the Kaldi source tree is `egs/wsj/s5`.

1. Prepare a speaker-transformed feature set given the feature transform specified in `final.feature_transform` and the feature files specified in `feats.scp`:

```sh
nnet-forward --use-gpu=no final.feature_transform "ark,s,cs:copy-feats scp:feats.scp ark:- |" ark:feat.ark
```

2. Score the feature set using the `speech_sample`:

```sh
./speech_sample -d GNA_AUTO -bs 8 -i feat.ark -m wsj_dnn5b.xml -o scores.ark
```
OpenVINO™ toolkit Intermediate Representation `wsj_dnn5b.xml` file was generated in the previous [Model Preparation](#model-preparation) section.

3. Run the Kaldi decoder to produce n-best text hypotheses and select most likely text given the WFST (`HCLG.fst`), vocabulary (`words.txt`), and TID/PID mapping (`final.mdl`):

```sh
latgen-faster-mapped --max-active=7000 --max-mem=50000000 --beam=13.0 --lattice-beam=6.0 --acoustic-scale=0.0833 --allow-partial=true --word-symbol-table=words.txt final.mdl HCLG.fst ark:scores.ark ark:-| lattice-scale --inv-acoustic-scale=13 ark:- ark:- | lattice-best-path --word-symbol-table=words.txt ark:- ark,t:-  > out.txt &
```

4. Run the word error rate tool to check accuracy given the vocabulary (`words.txt`) and reference transcript (`test_filt.txt`):

```sh
cat out.txt | utils/int2sym.pl -f 2- words.txt | sed s:\<UNK\>::g | compute-wer --text --mode=present ark:test_filt.txt ark,p:-
```

All of mentioned files can be downloaded from [https://storage.openvinotoolkit.org/models_contrib/speech/2021.2/wsj_dnn5b_smbr](https://storage.openvinotoolkit.org/models_contrib/speech/2021.2/wsj_dnn5b_smbr)

## See Also

- [Integrate the OpenVINO™ Runtime with Your Application](../../../docs/OV_Runtime_UG/Integrate_with_customer_application_new_API.md)
- [Using OpenVINO™ Toolkit Samples](../../../docs/OV_Runtime_UG/Samples_Overview.md)
- [Model Downloader](@ref omz_tools_downloader)
- [Model Optimizer](../../../docs/MO_DG/Deep_Learning_Model_Optimizer_DevGuide.md)