NVIDIA Audio Effects (AFX) SDK User Guide

https://docs.nvidia.com/maxine/afx/
Admin
The NVIDIA Audio Effects SDK provides various audio effects for broadcast use cases with real-time...
Tokens:42,740
Snippets:528
Trust Score:10
Update:3 months ago
Show doc for...
Tokens:
Raw
### Example Direct Run Command

Source: https://docs.nvidia.com/maxine/afx/latest/LinuxAFXSDK/SampleApplicationsLinux/EffectsDelayedStreamsDemoApplication.html

An example of directly running the sample application with a specific configuration file.

```bash
build/effects_delayed_streams_demo -c t4_denoise48k_10_cfg.txt

```

--------------------------------

### Example Directory Structure After Installation

Source: https://docs.nvidia.com/maxine/afx/latest/LinuxAFXSDK/InstallTheAFXSDK.html

This illustrates the expected folder structure within the Audio_Effects_SDK/features directory after a feature package, such as denoiser, has been successfully installed.

```tree
$ tree Audio_Effects_SDK/features/
Audio_Effects_SDK/features/
├── README.md
├── denoiser
│   ├── include
│   │   └── denoiser.h
│   ├── lib
│   │   ├── libnv_audiofx_denoiser.so -> libnv_audiofx_denoiser.so.2
│   │   ├── libnv_audiofx_denoiser.so.2 -> libnv_audiofx_denoiser.so.2.0.0
│   │   └── libnv_audiofx_denoiser.so.2.0.0
│   └── models
│       └── sm_89
│           ├── denoiser_16k.trtpkg
│           ├── denoiser_16k_6656.trtpkg
│           ├── denoiser_48k.trtpkg
│           ├── denoiser_48k_6656.trtpkg
│           ├── denoiser_v2_16k.trtpkg
│           ├── denoiser_v2_16k_7680.trtpkg
│           ├── denoiser_v2_48k.trtpkg
│           └── denoiser_v2_48k_3840.trtpkg


```

--------------------------------

### Run Noise Removal Sample App on Windows

Source: https://docs.nvidia.com/maxine/afx/latest/AboutTheEffects/AboutNoiseRemovalBackgroundNoiseSuppression.html

Initial setup to download models and then run the denoiser effect demo for 16k and 48k sample rates on Windows.

```bash
# (One time, initial setup): Download models using models/download_models.ps1
powershell -ExecutionPolicy Bypass -File ./download_models.ps1 --gpu_architecture <gpu> --effects denoiser-16k,denoiser-48k

# Format: run_effect_demo.bat <architecture> <effect> <input_sample_rate> <output_sample_rate>

# 16k effect
run_effect_demo.bat turing denoiser 16k 16k

# 48k effect
run_effect_demo.bat ampere denoiser 48k 48k
```

--------------------------------

### Run Noise Removal Sample App on Linux

Source: https://docs.nvidia.com/maxine/afx/latest/AboutTheEffects/AboutNoiseRemovalBackgroundNoiseSuppression.html

Initial setup to download models and then run the denoiser effect demo for 16k and 48k sample rates on Linux.

```bash
# (One time, initial setup): Download models using models/download_models.sh
./download_models.sh --gpu <gpu> --effects denoiser-16k,denoiser-48k

# Refer to Section 3.2 for further details
Format: ./run_effect.sh -g <gpu> -s <sample_rate> -e denoiser

# 16k effect
./run_effect.sh -g t4 -s 16 -e denoiser

# 48k effect
./run_effect.sh -g t4 -s 48 -e denoiser
```

--------------------------------

### Download Models for Speaker Focus (Windows)

Source: https://docs.nvidia.com/maxine/afx/latest/AboutTheEffects/AboutSpeakerFocusEffect.html

Initial setup command to download necessary models for the Speaker Focus effect on Windows. Specify your GPU architecture.

```bash
# (One time, initial setup): Download models using models/download_models.ps1
powershell -ExecutionPolicy Bypass -File ./download_models.ps1 --gpu_architecture <gpu> --effects speaker_focus-16k,speaker_focus-48k
```

--------------------------------

### Download Models for Speaker Focus (Linux)

Source: https://docs.nvidia.com/maxine/afx/latest/AboutTheEffects/AboutSpeakerFocusEffect.html

Initial setup command to download necessary models for the Speaker Focus effect on Linux. Specify your GPU architecture.

```bash
# (One time, initial setup): Download models using models/download_models.sh
./download_models.sh --gpu <gpu> --effects speaker_focus-16k,speaker_focus-48k
```

--------------------------------

### Input WAV List Configuration

Source: https://docs.nvidia.com/maxine/afx/latest/LinuxAFXSDK/SampleApplicationsLinux/EffectsDemoApplication.html

Example of specifying multiple input WAV files for different streams and sequential processing within a stream.

```plaintext
input_wav_list file1.wav file2.wav file3.wav;file4.wav;file5.wav file6.wav
```

--------------------------------

### Download Models and Run Demo (Windows)

Source: https://docs.nvidia.com/maxine/afx/latest/AboutTheEffects/AboutRoomEchoRemovalCancellation.html

Initial setup to download necessary models and then run the dereverb effect demo for 16k and 48k sample rates on Windows.

```bash
# (One time, initial setup): Download models using models/download_models.ps1
powershell -ExecutionPolicy Bypass -File ./download_models.ps1 --gpu_architecture <gpu> --effects dereverb-16k,dereverb-48k

# Format: run_effect_demo.bat <architecture> <effect> <input_sample_rate> <output_sample_rate>

# 16k effect
run_effect_demo.bat turing dereverb 16k 16k

# 48k effect
run_effect_demo.bat ampere dereverb 48k 48k
```

--------------------------------

### Download and Run Noise Removal/Room Echo Removal on Windows

Source: https://docs.nvidia.com/maxine/afx/latest/AboutTheEffects/AboutNoiseRemovalRoomEchoRemoval.html

Initial setup involves downloading models. Then, run the effect demo using the provided batch script, specifying the GPU architecture, effect name, and sample rates.

```bash
# (One time, initial setup): Download models using models/download_models.ps1
powershell -ExecutionPolicy Bypass -File ./download_models.ps1 --gpu_architecture <gpu> --effects dereverb_denoiser-16k,dereverb_denoiser-48k

# Format: run_effect_demo.bat <architecture> <effect> <input_sample_rate> <output_sample_rate>

# 16k effect
run_effect_demo.bat turing dereverb_denoiser 16k 16k

# 48k effect
run_effect_demo.bat ampere dereverb_denoiser 48k 48k

```

--------------------------------

### Download Models for Studio Voice Effect (Linux)

Source: https://docs.nvidia.com/maxine/afx/latest/AboutTheEffects/AboutStudioVoiceEffect.html

One-time setup to download necessary models for the Studio Voice effect on Linux. Specify GPU and effects.

```bash
# (One time, initial setup): Download models using models/download_models.sh
./download_models.sh --gpu <gpu> --effects studio_voice-16k,studio_voice-48k
```

--------------------------------

### Get Help for Download Script

Source: https://docs.nvidia.com/maxine/afx/latest/WindowsAFXSDK/InstallTheAFXSDK.html

Run the download script with the '--help' flag to view all available options and parameters for downloading AFX SDK features.

```powershell
powershell -ExecutionPolicy Bypass -File ./download_features.ps1 --help
```

--------------------------------

### Download Models for Voice Font Effect (Linux)

Source: https://docs.nvidia.com/maxine/afx/latest/AboutTheEffects/AboutVoiceFontEffect.html

Run this one-time setup command on Linux to download the necessary models for the voice font effect. Replace `<gpu>` with your GPU architecture.

```bash
./download_models.sh --gpu <gpu> --effects voice_font
```

--------------------------------

### Download Models and Run Demo (Linux)

Source: https://docs.nvidia.com/maxine/afx/latest/AboutTheEffects/AboutRoomEchoRemovalCancellation.html

Initial setup to download necessary models and then run the dereverb effect demo for 16k and 48k sample rates on Linux.

```bash
# (One time, initial setup): Download models using models/download_models.sh
./download_models.sh --gpu <gpu> --effects dereverb-16k,dereverb-48k

# Refer to Section 3.2 for further details
Format: ./run_effect.sh -g <gpu> -s <sample_rate> -e dereverb

# 16k effect
./run_effect.sh -g t4 -s 16 -e dereverb

# 48k effect
./run_effect.sh -g t4 -s 48 -e dereverb
```

--------------------------------

### Download Models for Voice Font Effect (Windows)

Source: https://docs.nvidia.com/maxine/afx/latest/AboutTheEffects/AboutVoiceFontEffect.html

Run this one-time setup command on Windows to download the necessary models for the voice font effect. Replace `<gpu>` with your GPU architecture.

```powershell
powershell -ExecutionPolicy Bypass -File ./download_models.ps1 --gpu_architecture <gpu> --effects voice_font
```

--------------------------------

### Setup and Run AEC Effect on Linux

Source: https://docs.nvidia.com/maxine/afx/latest/AboutTheEffects/AboutAcousticEchoCancellation.html

Initial setup for running the AEC effect on Linux, including downloading models and executing the effect script. Supports 16kHz and 48kHz effects with specified GPU and sample rate.

```bash
# (One time, initial setup): Download models using models/download_models.sh
./download_models.sh --gpu <gpu> --effects aec-16k,aec-48k

# Refer to Section 3.2 for further details
Format: ./run_effect.sh -g <gpu> -s <sample_rate> -e aec

# 16k effect
./run_effect.sh -g t4 -s 16 -e aec

# 48k effect
./run_effect.sh -g t4 -s 48 -e aec

```

--------------------------------

### Download Models for Studio Voice Effect (Windows)

Source: https://docs.nvidia.com/maxine/afx/latest/AboutTheEffects/AboutStudioVoiceEffect.html

One-time setup to download necessary models for the Studio Voice effect on Windows. Specify your GPU architecture.

```powershell
# (One time, initial setup): Download models using models/download_models.ps1
powershell -ExecutionPolicy Bypass -File ./download_models.ps1 --gpu_architecture <gpu> --effects studio_voice-48k
```

--------------------------------

### Install CUDA compatibility package

Source: https://docs.nvidia.com/maxine/afx/latest/LinuxAFXSDK/InstallTheAFXSDK.html

Installs the necessary compatibility libraries for older CUDA drivers. These libraries are typically installed under `/usr/local/cuda-12.8/compat`.

```bash
$ sudo apt-get install -y cuda-compat-12-8

```

--------------------------------

### Example: Run 16-kHz AEC Effect on Ada GPU

Source: https://docs.nvidia.com/maxine/afx/latest/WindowsAFXSDK/RunTheSampleApplication.html

This example demonstrates how to run the 16-kHz Acoustic Echo Cancellation (AEC) effect on an Ada Lovelace GPU architecture using the helper script.

```batch
run_effects_demo.bat -g ada -e aec -isr 16k -osr 16k
```

--------------------------------

### Run AFX SDK Container

Source: https://docs.nvidia.com/maxine/afx/latest/LinuxAFXSDK/UseSDKInContainers.html

This command starts a new Docker container, mapping a local output folder to '/host' inside the container. Ensure the output folder path does not end with a '/'.

```bash
./run_container_image.sh -m out

```

--------------------------------

### Run Effect Sample in Container

Source: https://docs.nvidia.com/maxine/afx/latest/LinuxAFXSDK/UseSDKInContainers.html

Executes the effects_demo sample within the AFX SDK container. This example runs the denoiser effect with a sample rate of 48kHz. The output is copied to the host-mapped directory '/host'.

```bash
cd samples/effects_demo/
./run_effect.sh -g a16 -e denoiser -s 48
cp -r denoiser/ /host/

```

--------------------------------

### Download Specific AFX SDK Features

Source: https://docs.nvidia.com/maxine/afx/latest/WindowsAFXSDK/InstallTheAFXSDK.html

Download only specific AI features by listing them with the '--effects' parameter. This example downloads Superresolution and AEC features for Turing architecture.

```powershell
powershell -ExecutionPolicy Bypass -File ./download_features.ps1 --gpu_architecture turing --ngc-org nvidia --ngc-team maxine --effects superres-16k_to_48k,superres-8k_to_16k,aec-16k,aec-48k
```

--------------------------------

### Get Audio Effect Configuration Parameters

Source: https://docs.nvidia.com/maxine/afx/latest/UseAFXInApps/GetParametersOfAnEffect.html

Retrieve the number of samples per input frame, the number of input channels, and the output sample rate for the audio effect.

```cpp
unsigned num_samples_per_frame, num_channels, sample_rate;
err = NvAFX_GetU32(handle, NVAFX_PARAM_NUM_SAMPLES_PER_INPUT_FRAME, &num_samples_per_frame);
err = NvAFX_GetU32(handle, NVAFX_PARAM_NUM_INPUT_CHANNELS, &num_channels);
err = NvAFX_GetU32(handle, NVAFX_PARAM_OUTPUT_SAMPLE_RATE, &sample_rate);
```

--------------------------------

### Query Supported Samples Per Frame

Source: https://docs.nvidia.com/maxine/afx/latest/UseAFXInApps/GetParametersOfAnEffect.html

Query the effect for the supported number of samples per frame. Initialize list_size to zero to get the actual size, then allocate memory and query again.

```cpp
NvAFX_Status err;

std::unique_ptr<unsigned int[]> supported_list = nullptr;
int list_size = 0;

// Query with list_size set to zero to get the actual list size
err = NvAFX_GetU32List(handle, NVAFX_PARAM_SUPPORTED_NUM_SAMPLES_PER_FRAME, supported_list.get(), &list_size);
if (err != NVAFX_STATUS_OUTPUT_BUFFER_TOO_SMALL) {
             // This indicates API failure
             return;
}

// Allocate memory based on the returned size, and query with this pointer
supported_list.reset(new unsigned int[list_size]);
err = NvAFX_GetU32List(handle,
NVAFX_PARAM_SUPPORTED_NUM_SAMPLES_PER_FRAME, supported_list.get(), &list_size);
```

--------------------------------

### Compile the Application

Source: https://docs.nvidia.com/maxine/afx/latest/LinuxAFXSDK/SampleApplicationsLinux/EffectsDelayedStreamsDemoApplication.html

Navigate to the sample directory and run the build script to compile the application.

```bash
:/Audio Effects SDK/samples/effects_delayed_streams_demo$ ./build.sh

```

--------------------------------

### Update apt repository cache

Source: https://docs.nvidia.com/maxine/afx/latest/LinuxAFXSDK/InstallTheAFXSDK.html

Run this command to update the package list before installing new packages.

```bash
$ sudo apt-get update

```

--------------------------------

### Run Sample Application Directly

Source: https://docs.nvidia.com/maxine/afx/latest/LinuxAFXSDK/SampleApplicationsLinux/EffectsDelayedStreamsDemoApplication.html

Execute the built sample application directly using its executable, specifying a configuration file.

```bash
build/effects_delayed_streams_demo -c <config-file>

```

--------------------------------

### Run Effects Demo Application

Source: https://docs.nvidia.com/maxine/afx/latest/LinuxAFXSDK/SampleApplicationsLinux/EffectsDemoApplication.html

Execute the built sample application with a specified configuration file.

```bash
build/effects_demo -c <config-file>
```

--------------------------------

### Get Number of Available GPUs

Source: https://docs.nvidia.com/maxine/afx/latest/UseAFXInApps/UseMultipleGPUs.html

Call `cudaGetDeviceCount()` to determine the number of GPUs available on the system. This is the first step in managing multiple GPUs.

```c++
cuErr = cudaGetDeviceCount(&deviceCount);

```

--------------------------------

### Get Supported Audio Devices

Source: https://docs.nvidia.com/maxine/afx/latest/APIReference/AFXFunctions.html

Retrieves a list of CUDA device indices compatible with the current model file. This function is not suitable for chained effects.

```c
NvAFX_Status NvAFX_GetSupportedDevices(
   NvAFX_Handle effect,
   int *num,
   int *devices
);

```

--------------------------------

### Run Sample Application with Helper Script

Source: https://docs.nvidia.com/maxine/afx/latest/LinuxAFXSDK/SampleApplicationsLinux/EffectsDelayedStreamsDemoApplication.html

Use the run_effect.sh helper script to execute the sample application with specified parameters, including device, sample rate, batch size, effect, and stream activation.

```bash
./run_effect.sh -g t4 -s 16 -b 10 -e denoiser -a

```

--------------------------------

### Run Effects Demo with Specific Config

Source: https://docs.nvidia.com/maxine/afx/latest/WindowsAFXSDK/RunTheSampleApplication.html

Execute the sample application with a specific configuration file for the denoiser effect, including the architecture. Replace <architecture> with the appropriate value.

```bash
effects_demo.exe -c denoise48k_cfg_<architecture>.txt
```

--------------------------------

### Run Audio Super-Resolution Demo on Linux

Source: https://docs.nvidia.com/maxine/afx/latest/AboutTheEffects/AboutAudioSuperResolution.html

Execute the sample application on Linux for the Audio Super-Resolution effect. Ensure models are downloaded first. Specify GPU, sample rate, and effect.

```bash
# (One time, initial setup): Download models using models/download_models.sh
./download_models.sh --gpu <gpu> --effects superres-8k_to_16k,superres-16k_to_48k

# Refer to Section 3.2 for further details
Format: ./run_effect.sh -g <gpu> -s <sample_rate> -e superres

# 16k effect
./run_effect.sh -g t4 -s 16 -e superres

# 48k effect
./run_effect.sh -g t4 -s 48 -e superres

```

--------------------------------

### Load AFX SDK DLL at Runtime (Windows)

Source: https://docs.nvidia.com/maxine/afx/latest/UseAFXInApps/BuildApplicationsWithTheSDK.html

Demonstrates how to dynamically load the AFX SDK DLL and access its functions using LoadLibrary and GetProcAddress. Ensure the DLL is in a location discoverable by the system.

```cpp
typedef NvAFX_Status(*NVAFX_CREATEEFFECT)(NvAFX_EffectSelector, NvAFX_Handle*);
HINSTANCE h = LoadLibraryW(L"NVAudioEffects.dll");
NvAFX_CreateEffect = (NVAFX_CREATEEFFECT)GetProcAddress(h, "NvAFX_CreateEffect");
void *nv_handle;
NvAFX_CreateEffect("denoiser", &nv_handle);

// Similarly for other APIs
FreeLibrary(h);

```

--------------------------------

### Run Audio Super-Resolution Demo on Windows

Source: https://docs.nvidia.com/maxine/afx/latest/AboutTheEffects/AboutAudioSuperResolution.html

Execute the sample application on Windows for the Audio Super-Resolution effect. Ensure models are downloaded first. Specify GPU architecture, effect name, and sample rates.

```batch
# (One time, initial setup): Download models using models/download_models.ps1
powershell -ExecutionPolicy Bypass -File ./download_models.ps1 --gpu_architecture <gpu> --effects superres-8k_to_16k,superres-16k_to_48k

# Format: run_effect_demo.bat <architecture> <effect> <input_sample_rate> <output_sample_rate>

# 8k - 16k effect
run_effect_demo.bat turing superres 8k 16k

# 16k - 48k effect
run_effect_demo.bat ampere superres 16k 48k

```

--------------------------------

### Run Effects Demo with Config File

Source: https://docs.nvidia.com/maxine/afx/latest/WindowsAFXSDK/RunTheSampleApplication.html

Execute the sample application by specifying a configuration file. The config file path is a placeholder that should be replaced with the actual path to your configuration file.

```bash
effects_demo.exe -c <config-file>
```

--------------------------------

### Download and Run Noise Removal/Room Echo Removal on Linux

Source: https://docs.nvidia.com/maxine/afx/latest/AboutTheEffects/AboutNoiseRemovalRoomEchoRemoval.html

First, download the necessary models using the provided shell script. Then, execute the effect demo script with the appropriate arguments for GPU, sample rate, and effect.

```bash
# (One time, initial setup): Download models using models/download_models.sh
./download_models.sh --gpu <gpu> --effects dereverb_denoiser-16k,dereverb_denoiser-48k

# Refer to Section 3.2 for further details
Format: ./run_effect.sh -g <gpu> -s <sample_rate> -e <dereverb_denoniser>

# 16k effect
./run_effect.sh -g t4 -s 16 -e dereverb_denoiser

# 48k effect
./run_effect.sh -g t4 -s 48 -e dereverb_denoiser

```

--------------------------------

### Dynamic Loading of SDK Library

Source: https://docs.nvidia.com/maxine/afx/latest/UseAFXInApps/BuildApplicationsWithTheSDK.html

Dynamically load the `libnv_audiofx.so` library at runtime using `dlopen` and `dlsym`. Ensure the library path is set correctly (e.g., via `LD_LIBRARY_PATH`).

```c++
// Typedefs for functions; similarly define for other functions required

typedef NvAFX_Status (*fnNvAFX_CreateEffect)(NvAFX_EffectSelector code, NvAFX_Handle* effect);

// Load library and bind

// Note: ensure that external/cuda/lib is in library path, or fix via RPATH/similar

void* handle = dlopen("libnv_audiofx.so", RTLD_LAZY);

assert(handle);

fnNvAFX_CreateEffect f = (fnNvAFX_CreateEffect) dlsym(handle, "NvAFX_CreateEffect");

assert(f);

// Call functions

NvAFX_Handle effect;

auto ret = f(NVAFX_EFFECT_DENOISER, &effect);

assert(ret == NVAFX_STATUS_SUCCESS);

```

--------------------------------

### Set Voice Font Reference Audio (Linux)

Source: https://docs.nvidia.com/maxine/afx/latest/UseAFXInApps/UseTheAFXSDKInApplications.html

Sets the reference audio for selected streams for the Voice Font effect on Linux. Stream IDs start at zero. Multiple calls overwrite previous settings.

```C
// Set the reference of selected streams

// ID starts at zero (first stream is zero)

// If this function is called multiple times, it overwrites the previously set reference

std::vector<unsigned int> stream_idx (num_streams) = {1,2};

std::vector<const float*> reference_frame_ptrs(num_streams) = getRefForStreams(stream_idx);

NvAFX_SetStreamFloatList(handle, NVAFX_PARAM_REFERENCE_AUDIO, stream_idx.data(), reference_frame_ptrs.data(), num_streams);
```

--------------------------------

### Run Chained Effects with Configuration File

Source: https://docs.nvidia.com/maxine/afx/latest/WindowsAFXSDK/ChainMultipleEffects.html

Execute the `effects_demo` application with a configuration file that specifies a chain of audio effects. This command demonstrates running the Superresolution effect followed by the Denoiser effect.

```bash
effects_demo.exe -c superres16kto48k_denoise48k_chain_cfg.txt
```

--------------------------------

### Create Chained Denoiser and Superresolution Effect

Source: https://docs.nvidia.com/maxine/afx/latest/UseAFXInApps/CreateChainedAudioEffect.html

This example demonstrates how to create a chained audio effect combining Background Noise Removal (16 kHz) and Superresolution (16 kHz to 48 kHz) using the `NvAFX_CreateChainedEffect` function.

```c
NvAFX_Status err = NvAFX_CreateChainedEffect(NVAFX_CHAINED_EFFECT_DENOISER_16k_SUPERRES_16k_TO_48k, &handle);
```

--------------------------------

### Set LD_LIBRARY_PATH for AFX SDK

Source: https://docs.nvidia.com/maxine/afx/latest/LinuxAFXSDK/SampleApplicationsLinux/EffectsDemoApplication.html

Export the LD_LIBRARY_PATH environment variable to include the AFX SDK's external CUDA library directory. This ensures the application loads compatible library versions and avoids conflicts with system-wide installations.

```bash
export LD_LIBRARY_PATH=external/cuda/lib:$LD_LIBRARY_PATH

```

--------------------------------

### Run Speaker Focus Effect Demo (Windows)

Source: https://docs.nvidia.com/maxine/afx/latest/AboutTheEffects/AboutSpeakerFocusEffect.html

Command to run the sample application for the Speaker Focus effect on Windows. Adjust parameters for different GPU architectures, sampling rates, and effect versions.

```bash
# Format: run_effects_demo.bat -g ^<architecture^> -e ^<effect^> -isr ^<input_sr^> -osr ^<output_sr^> -ir ^<intensity_ratio^> -ev ^<effect_version^> -vad ^<enable_vad^>

# 16k effect on turing GPU
run_effects_demo.bat -g turing -e speaker_focus -isr 16k -osr 16k

# 48k effect on ampere GPU
run_effects_demo.bat -g ampere -e speaker_focus -isr 48k -osr 48k
```

--------------------------------

### Run Effects Demo with Helper Script

Source: https://docs.nvidia.com/maxine/afx/latest/LinuxAFXSDK/SampleApplicationsLinux/EffectsDemoApplication.html

Use the `run_effect.sh` script to execute the effects_demo application. This script generates a configuration file based on the provided arguments and then runs the demo. Specify GPU, effect, sample rates, batch size, and input files/folders.

```bash
./run_effect.sh -g <gpu> -e <effect> -s <input_sample_rate> -o <output_sample_rate> -b <batch_size> -i <input_file_or_folder>
```

```bash
./run_effect.sh -g t4 -s 16 -b 10 -e denoiser
```

--------------------------------

### Fetch Supported GPU Devices on Windows

Source: https://docs.nvidia.com/maxine/afx/latest/UseAFXInApps/GetSupportedDevicesOnWindows.html

Call NvAFX_GetSupportedDevices twice: first with a null buffer to get the count, then with a vector to populate it with supported CUDA device indices. This method is not supported for chained effects and must be called after setting the model path.

```cpp
int numSupportedDevices = 0;
NvAFX_GetSupportedDevices(handle, &numSupportedDevices, nullptr);
std::vector<int> ret(num);
NvAFX_GetSupportedDevices(handle, &numSupportedDevices, ret.data);
```

--------------------------------

### Run BNR 2.0 on Linux

Source: https://docs.nvidia.com/maxine/afx/latest/AboutTheEffects/AboutNoiseRemovalBackgroundNoiseSuppression.html

Execute the sample application for BNR 2.0 on Linux. Ensure models are downloaded first. This command runs the 16k effect.

```bash
# (One time, initial setup): Download models using models/download_models.sh
./download_models.sh --gpu <gpu> --effects denoiser-16k,denoiser-48k

# Refer to Section 3.2 for further details
Format: ./run_effect.sh -g <gpu> -s <sample_rate> -e denoiser -m 2

# 16k effect
./run_effect.sh -g t4 -s 16 -e denoiser -m 2

# 48k effect
./run_effect.sh -g t4 -s 48 -e denoiser -m 2
```

--------------------------------

### NvAFX_InitializeLogger

Source: https://docs.nvidia.com/maxine/afx/latest/AFXSDKAPIReference.html

Initializes the SDK's logging system.

```APIDOC
## NvAFX_InitializeLogger

### Description
Initializes the logging system for the NVIDIA AFX SDK. This should be called before any other SDK functions.

### Function Signature
```c
NvAFX_API NvAFX_InitializeLogger(Logging_cb_t callback, LoggingSeverity minSeverity, LoggingTarget target);
```

### Parameters
#### Input Parameters
- **callback** (Logging_cb_t) - A callback function to receive log messages.
- **minSeverity** (LoggingSeverity) - The minimum severity level of messages to log.
- **target** (LoggingTarget) - The target for log messages (e.g., console, file).
```

--------------------------------

### Run Denoiser Sample with Architecture

Source: https://docs.nvidia.com/maxine/afx/latest/WindowsAFXSDK/RunTheSampleApplication.html

Execute the denoiser sample application using a batch script, specifying the target architecture. Replace <architecture> with the appropriate value (e.g., turing, ampere).

```batch
run_denoiser_48k.bat -g <architecture>
```

--------------------------------

### NvAFX_GetSupportedDevices

Source: https://docs.nvidia.com/maxine/afx/latest/AFXSDKAPIReference.html

Retrieves a list of supported audio devices (Windows SDK only).

```APIDOC
## NvAFX_GetSupportedDevices (Windows SDK only)

### Description
Retrieves a list of audio devices supported by the NVIDIA AFX SDK on Windows.

### Function Signature
```c
NvAFX_API NvAFX_GetSupportedDevices(NvAFX_EffectSelector* pDeviceList, uint32_t* pNumDevices);
```

### Parameters
#### Output Parameters
- **pDeviceList** (NvAFX_EffectSelector*) - Pointer to an array of NvAFX_EffectSelector structures to be filled with device information.
- **pNumDevices** (uint32_t*) - Pointer to a uint32_t that will be updated with the number of supported devices found.
```

--------------------------------

### Run Studio Voice High Quality Effect on Windows

Source: https://docs.nvidia.com/maxine/afx/latest/AboutTheEffects/AboutStudioVoiceEffect.html

Use this command to run the sample application for the Studio Voice High Quality effect on Windows. Ensure models are downloaded first.

```bash
# (One time, initial setup): Download models using models/download_models.ps1
powershell -ExecutionPolicy Bypass -File ./download_models.ps1 --gpu_architecture <gpu> --effects studio_voice-16k,studio_voice-48k

# Format: run_effects_demo.bat -g ^<architecture^> -e ^<effect^> -isr ^<input_sr^> -osr ^<output_sr^> -ir ^<intensity_ratio^> -ev ^<effect_version^> -vad ^<enable_vad^>

# 16k effect on turing GPU
run_effects_demo.bat -g turing -e studio_voice_high_quality -isr 16k -osr 16k

# 48k effect on ampere GPU
run_effects_demo.bat -g ampere -e studio_voice_high_quality -isr 48k -osr 48k

```

--------------------------------

### Compile-time Linking with GCC

Source: https://docs.nvidia.com/maxine/afx/latest/UseAFXInApps/BuildApplicationsWithTheSDK.html

Use this method to link the core SDK library `libnv_audiofx.so` during the compilation process. Ensure the library and include paths are correctly specified.

```bash
gcc -L "../../nvafx/lib" -l "nv_audiofx" -L "../../external/cuda/lib/" -I "../../nvafx/include" source.c
```

--------------------------------

### Download All Feature Packages Including Early Access

Source: https://docs.nvidia.com/maxine/afx/latest/LinuxAFXSDK/InstallTheAFXSDK.html

This command downloads all feature packages, including those in Early Access, for the automatically detected GPU. Ensure your NGC API key is set.

```bash
./download_features.sh --effects aec-16k,aec-48k,dereverb-16k,dereverb-48k,dereverb_denoiser-16k,dereverb_denoiser-48k,denoiser-16k,denoiser-48k,superres-8k_to_16k,superres-16k_to_48k,studio_voice-16k,studio_voice-48k,speaker_focus-16k,speaker_focus-48k,voice_font

```

--------------------------------

### Run Studio Voice High Quality Effect on Linux

Source: https://docs.nvidia.com/maxine/afx/latest/AboutTheEffects/AboutStudioVoiceEffect.html

Use this command to run the sample application for the Studio Voice High Quality effect on Linux. Ensure models are downloaded first.

```bash
# (One time, initial setup): Download models using models/download_models.sh
./download_models.sh --gpu <gpu> --effects studio_voice-16k,studio_voice-48k

# Refer to Section 3.2 for further details
Format: ./run_effect.sh -g <gpu> -s <sample_rate> -e studio_voice_high_quality

# 16k effect
./run_effect.sh -g t4 -s 16 -e studio_voice_high_quality

# 48k effect
./run_effect.sh -g t4 -s 48 -e studio_voice_high_quality

```

--------------------------------

### Run Studio Voice Low Latency Effect (Linux)

Source: https://docs.nvidia.com/maxine/afx/latest/AboutTheEffects/AboutStudioVoiceEffect.html

Execute the sample application for the Studio Voice Low Latency effect on Linux. Specify GPU, sample rate, and effect name.

```bash
# Refer to Section 3.2 for further details
Format: ./run_effect.sh -g <gpu> -s <sample_rate> -e studio_voice_low_latency

# 16k effect
./run_effect.sh -g t4 -s 16 -e studio_voice_low_latency

# 48k effect
./run_effect.sh -g t4 -s 48 -e studio_voice_low_latency
```

--------------------------------

### Run Studio Voice Low Latency Effect Demo (Windows)

Source: https://docs.nvidia.com/maxine/afx/latest/AboutTheEffects/AboutStudioVoiceEffect.html

Execute the sample application for the Studio Voice Low Latency effect on Windows. Configure GPU architecture, effect name, sample rates, intensity ratio, effect version, and VAD.

```batch
# Format: run_effects_demo.bat -g ^<architecture^> -e ^<effect^> -isr ^<input_sr^> -osr ^<output_sr^> -ir ^<intensity_ratio^> -ev ^<effect_version^> -vad ^<enable_vad^>

# 16k effect on turing GPU
run_effects_demo.bat -g turing -e studio_voice_low_latency -isr 16k -osr 16k

# 48k effect on ampere GPU
run_effects_demo.bat -g ampere -e studio_voice_low_latency -isr 48k -osr 48k
```

--------------------------------

### General Command for run_effects_demo.bat

Source: https://docs.nvidia.com/maxine/afx/latest/WindowsAFXSDK/RunTheSampleApplication.html

This is the general syntax for running the `run_effects_demo.bat` script. It allows specifying GPU architecture, effect, input/output sample rates, intensity ratio, effect version, and VAD enablement.

```batch
run_effects_demo.bat -g <architecture> -e <effect> -isr <input_sr> -osr <output_sr> -ir <intensity_ratio> -ev <effect_version> -vad <enable_vad>
```

--------------------------------

### Run Effects Demo with Auto GPU Detection

Source: https://docs.nvidia.com/maxine/afx/latest/LinuxAFXSDK/SampleApplicationsLinux/EffectsDemoApplication.html

Execute the effects demo script to automatically detect the current GPU and load appropriate models for various audio effects. This is useful for quick testing and demonstration.

```bash
./run_effect.sh -s 16 -b 1 -e denoiser
./run_effect.sh -s 48 -b 1 -e dereverb
./run_effect.sh -s 16 -b 400 -e denoiser
./run_effect.sh -s 48 -b 400 -e dereverb_denoiser
./run_effect.sh -s 48 -b 400 -e aec
./run_effect.sh -s 8 -o 16 -b 400 -e superres
./run_effect.sh -s 16 -b 1 -e studio_voice_high_quality
./run_effect.sh -s 48 -b 1 -e studio_voice_high_quality
./run_effect.sh -s 48 -b 1 -e studio_voice_low_latency
./run_effect.sh -s 16 -b 1 -e speaker_focus
./run_effect.sh -s 48 -b 1 -e speaker_focus
./run_effect.sh -s 16 -b 3 -e voice_font_high_quality
./run_effect.sh -s 48 -b 3 -e voice_font_high_quality
./run_effect.sh -s 16 -b 3 -e voice_font_low_latency
./run_effect.sh -s 48 -b 3 -e voice_font_low_latency

```

--------------------------------

### Download All Features for Detected GPU

Source: https://docs.nvidia.com/maxine/afx/latest/WindowsAFXSDK/InstallTheAFXSDK.html

If '--gpu_architecture' is not specified, the script attempts to detect the current GPU and downloads all features (excluding Early Access) for it.

```powershell
powershell -ExecutionPolicy Bypass -File ./download_features.ps1 --effects "speaker_focus-16k,speaker_focus-48k,voice_font_low_latency,voice_font_high_quality"
```

--------------------------------

### Chaining Two Effects with Helper Script

Source: https://docs.nvidia.com/maxine/afx/latest/WindowsAFXSDK/ChainMultipleEffects.html

Use this command to apply two effects sequentially on audio files. Specify the GPU architecture, the first and second effects, and their respective input and output sample rates.

```bash
run_effects_demo.bat -g <architecture> -e <effect_1> -isr <in_sr_1> -osr <out_sr_1> -e2 <effect_2> -isr2 <in_sr_2> -osr2 <out_sr_2>
```

--------------------------------

### Generate Configuration File

Source: https://docs.nvidia.com/maxine/afx/latest/LinuxAFXSDK/SampleApplicationsLinux/EffectsDemoApplication.html

Use the run_effect.sh script to generate a configuration file for a specific effect and sample rate.

```bash
./run_effect.sh -e aec -s 48 -g t4 -c t4_aec.cfg
```

--------------------------------

### Download All Feature Packages (Auto-Detected GPU)

Source: https://docs.nvidia.com/maxine/afx/latest/LinuxAFXSDK/InstallTheAFXSDK.html

This command downloads all available feature packages for the GPU detected on the system, excluding Early Access features. Ensure you have set your NGC API key first.

```bash
./download_features.sh

```

--------------------------------

### Chaining Multiple Effects

Source: https://docs.nvidia.com/maxine/afx/latest/LinuxAFXSDK/SampleApplicationsLinux/EffectsDemoApplication.html

Demonstrates how to chain multiple audio effects using a shell script. This script generates a configuration file for running effects in a sequence.

```bash
./run_effect_chained.sh -g <gpu> -e1 <effect1> -s1 <input_sample_rate_1> -o1 <output_sample_rate_1> -e2 <effect2> -s2 *input_sample_rate_2 -o2 output_sample_rate_2* [-c *path_to_save_config_file*] [-i input_file_or_folder]
```
Tokens:
Raw