使用语音 SDK 选择音频输入设备

本文介绍如何获取连接到系统的音频设备的 ID。 然后,可以在语音 SDK 中使用这些 ID 来选择音频输入。 通过 AudioConfig 对象配置音频设备:

audioConfig = AudioConfig.FromMicrophoneInput("<device id>");
audioConfig = AudioConfig.FromMicrophoneInput("<device id>");
audio_config = AudioConfig(device_name="<device id>");
audioConfig = AudioConfiguration.FromMicrophoneInput("<device id>");
audioConfig = AudioConfiguration.fromMicrophoneInput("<device id>");
audioConfig = AudioConfiguration.fromMicrophoneInput("<device id>");

注意

在 Node.js 中运行的 JavaScript 无法使用麦克风。

Windows 桌面版应用程序上的音频设备 ID

可以从 Windows 桌面版应用程序中的 IMMDevice 对象中检索音频设备终结点 ID 字符串

下面的代码示例说明了如何使用它在 C++ 中枚举音频设备:

#include <cstdio>
#include <mmdeviceapi.h>

#include <Functiondiscoverykeys_devpkey.h>

const CLSID CLSID_MMDeviceEnumerator = __uuidof(MMDeviceEnumerator);
const IID IID_IMMDeviceEnumerator = __uuidof(IMMDeviceEnumerator);

constexpr auto REFTIMES_PER_SEC = (10000000 * 25);
constexpr auto REFTIMES_PER_MILLISEC = 10000;

#define EXIT_ON_ERROR(hres)  \
              if (FAILED(hres)) { goto Exit; }
#define SAFE_RELEASE(punk)  \
              if ((punk) != NULL)  \
                { (punk)->Release(); (punk) = NULL; }

void ListEndpoints();

int main()
{
    CoInitializeEx(NULL, COINIT_MULTITHREADED);
    ListEndpoints();
}

//-----------------------------------------------------------
// This function enumerates all active (plugged in) audio
// rendering endpoint devices. It prints the friendly name
// and endpoint ID string of each endpoint device.
//-----------------------------------------------------------
void ListEndpoints()
{
    HRESULT hr = S_OK;
    IMMDeviceEnumerator *pEnumerator = NULL;
    IMMDeviceCollection *pCollection = NULL;
    IMMDevice *pEndpoint = NULL;
    IPropertyStore *pProps = NULL;
    LPWSTR pwszID = NULL;

    hr = CoCreateInstance(CLSID_MMDeviceEnumerator, NULL, CLSCTX_ALL, IID_IMMDeviceEnumerator, (void**)&pEnumerator);
    EXIT_ON_ERROR(hr);

    hr = pEnumerator->EnumAudioEndpoints(eCapture, DEVICE_STATE_ACTIVE, &pCollection);
    EXIT_ON_ERROR(hr);

    UINT  count;
    hr = pCollection->GetCount(&count);
    EXIT_ON_ERROR(hr);

    if (count == 0)
    {
        printf("No endpoints found.\n");
    }

    // Each iteration prints the name of an endpoint device.
    PROPVARIANT varName;
    for (ULONG i = 0; i < count; i++)
    {
        // Get the pointer to endpoint number i.
        hr = pCollection->Item(i, &pEndpoint);
        EXIT_ON_ERROR(hr);

        // Get the endpoint ID string.
        hr = pEndpoint->GetId(&pwszID);
        EXIT_ON_ERROR(hr);

        hr = pEndpoint->OpenPropertyStore(
            STGM_READ, &pProps);
        EXIT_ON_ERROR(hr);

        // Initialize the container for property value.
        PropVariantInit(&varName);

        // Get the endpoint's friendly-name property.
        hr = pProps->GetValue(PKEY_Device_FriendlyName, &varName);
        EXIT_ON_ERROR(hr);

        // Print the endpoint friendly name and endpoint ID.
        printf("Endpoint %d: \"%S\" (%S)\n", i, varName.pwszVal, pwszID);

        CoTaskMemFree(pwszID);
        pwszID = NULL;
        PropVariantClear(&varName);
    }

Exit:
    CoTaskMemFree(pwszID);
    pwszID = NULL;
    PropVariantClear(&varName);
    SAFE_RELEASE(pEnumerator);
    SAFE_RELEASE(pCollection);
    SAFE_RELEASE(pEndpoint);
    SAFE_RELEASE(pProps);
}

在 C# 中,可以使用 NAudio 库来访问 CoreAudio API 并枚举设备,如下所示:

using System;

using NAudio.CoreAudioApi;

namespace ConsoleApp
{
    class Program
    {
        static void Main(string[] args)
        {
            var enumerator = new MMDeviceEnumerator();
            foreach (var endpoint in
                     enumerator.EnumerateAudioEndPoints(DataFlow.Capture, DeviceState.Active))
            {
                Console.WriteLine("{0} ({1})", endpoint.FriendlyName, endpoint.ID);
            }
        }
    }
}

示例设备 ID 为 {0.0.1.00000000}.{5f23ab69-6181-4f4a-81a4-45414013aac8}

UWP 上的音频设备 ID

在通用 Windows 平台 (UWP) 上,可以使用相应 DeviceInformation 对象的 Id() 属性来获取音频输入设备。

以下代码示例演示了如何在 C++ 和 C# 中执行此步骤:

#include <winrt/Windows.Foundation.h>
#include <winrt/Windows.Devices.Enumeration.h>

using namespace winrt::Windows::Devices::Enumeration;

void enumerateDeviceIds()
{
    auto promise = DeviceInformation::FindAllAsync(DeviceClass::AudioCapture);

    promise.Completed(
        [](winrt::Windows::Foundation::IAsyncOperation<DeviceInformationCollection> const& sender,
           winrt::Windows::Foundation::AsyncStatus /* asyncStatus */) {
        auto info = sender.GetResults();
        auto num_devices = info.Size();

        for (const auto &device : info)
        {
            std::wstringstream ss{};
            ss << "looking at device (of " << num_devices << "): " << device.Id().c_str() << "\n";
            OutputDebugString(ss.str().c_str());
        }
    });
}
using Windows.Devices.Enumeration;
using System.Linq;

namespace helloworld {
    private async void EnumerateDevices()
    {
        var devices = await DeviceInformation.FindAllAsync(DeviceClass.AudioCapture);

        foreach (var device in devices)
        {
            Console.WriteLine($"{device.Name}, {device.Id}\n");
        }
    }
}

示例设备 ID 为 \\\\?\\SWD#MMDEVAPI#{0.0.1.00000000}.{5f23ab69-6181-4f4a-81a4-45414013aac8}#{2eef81be-33fa-4800-9670-1cd474972c3f}

Linux 上的音频设备 ID

使用标准 ALSA 设备 ID 选择设备 ID。

附加到系统的输入的 ID 包含在命令 arecord -L 的输出中。 也可使用 ALSA C 库来获取它们。

示例 ID 为 hw:1,0hw:CARD=CC,DEV=0

macOS 上的音频设备 ID

在 Objective-C 中实现的以下函数可创建一个列表,其中包含附加到 Mac 的音频设备的名称和 ID。

deviceUID 字符串用于标识 macOS 版语音 SDK 中的设备。

#import <Foundation/Foundation.h>
#import <CoreAudio/CoreAudio.h>

CFArrayRef CreateInputDeviceArray()
{
    AudioObjectPropertyAddress propertyAddress = {
        kAudioHardwarePropertyDevices,
        kAudioObjectPropertyScopeGlobal,
        kAudioObjectPropertyElementMaster
    };

    UInt32 dataSize = 0;
    OSStatus status = AudioObjectGetPropertyDataSize(kAudioObjectSystemObject, &propertyAddress, 0, NULL, &dataSize);
    if (kAudioHardwareNoError != status) {
        fprintf(stderr, "AudioObjectGetPropertyDataSize (kAudioHardwarePropertyDevices) failed: %i\n", status);
        return NULL;
    }

    UInt32 deviceCount = (uint32)(dataSize / sizeof(AudioDeviceID));

    AudioDeviceID *audioDevices = (AudioDeviceID *)(malloc(dataSize));
    if (NULL == audioDevices) {
        fputs("Unable to allocate memory", stderr);
        return NULL;
    }

    status = AudioObjectGetPropertyData(kAudioObjectSystemObject, &propertyAddress, 0, NULL, &dataSize, audioDevices);
    if (kAudioHardwareNoError != status) {
        fprintf(stderr, "AudioObjectGetPropertyData (kAudioHardwarePropertyDevices) failed: %i\n", status);
        free(audioDevices);
        audioDevices = NULL;
        return NULL;
    }

    CFMutableArrayRef inputDeviceArray = CFArrayCreateMutable(kCFAllocatorDefault, deviceCount, &kCFTypeArrayCallBacks);
    if (NULL == inputDeviceArray) {
        fputs("CFArrayCreateMutable failed", stderr);
        free(audioDevices);
        audioDevices = NULL;
        return NULL;
    }

    // Iterate through all the devices and determine which are input-capable
    propertyAddress.mScope = kAudioDevicePropertyScopeInput;
    for (UInt32 i = 0; i < deviceCount; ++i) {
        // Query device UID
        CFStringRef deviceUID = NULL;
        dataSize = sizeof(deviceUID);
        propertyAddress.mSelector = kAudioDevicePropertyDeviceUID;
        status = AudioObjectGetPropertyData(audioDevices[i], &propertyAddress, 0, NULL, &dataSize, &deviceUID);
        if (kAudioHardwareNoError != status) {
            fprintf(stderr, "AudioObjectGetPropertyData (kAudioDevicePropertyDeviceUID) failed: %i\n", status);
            continue;
        }

        // Query device name
        CFStringRef deviceName = NULL;
        dataSize = sizeof(deviceName);
        propertyAddress.mSelector = kAudioDevicePropertyDeviceNameCFString;
        status = AudioObjectGetPropertyData(audioDevices[i], &propertyAddress, 0, NULL, &dataSize, &deviceName);
        if (kAudioHardwareNoError != status) {
            fprintf(stderr, "AudioObjectGetPropertyData (kAudioDevicePropertyDeviceNameCFString) failed: %i\n", status);
            continue;
        }

        // Determine if the device is an input device (it is an input device if it has input channels)
        dataSize = 0;
        propertyAddress.mSelector = kAudioDevicePropertyStreamConfiguration;
        status = AudioObjectGetPropertyDataSize(audioDevices[i], &propertyAddress, 0, NULL, &dataSize);
        if (kAudioHardwareNoError != status) {
            fprintf(stderr, "AudioObjectGetPropertyDataSize (kAudioDevicePropertyStreamConfiguration) failed: %i\n", status);
            continue;
        }

        AudioBufferList *bufferList = (AudioBufferList *)(malloc(dataSize));
        if (NULL == bufferList) {
            fputs("Unable to allocate memory", stderr);
            break;
        }

        status = AudioObjectGetPropertyData(audioDevices[i], &propertyAddress, 0, NULL, &dataSize, bufferList);
        if (kAudioHardwareNoError != status || 0 == bufferList->mNumberBuffers) {
            if (kAudioHardwareNoError != status)
                fprintf(stderr, "AudioObjectGetPropertyData (kAudioDevicePropertyStreamConfiguration) failed: %i\n", status);
            free(bufferList);
            bufferList = NULL;
            continue;
        }

        free(bufferList);
        bufferList = NULL;

        // Add a dictionary for this device to the array of input devices
        CFStringRef keys    []  = { CFSTR("deviceUID"),     CFSTR("deviceName")};
        CFStringRef values  []  = { deviceUID,              deviceName};

        CFDictionaryRef deviceDictionary = CFDictionaryCreate(kCFAllocatorDefault,
                                                              (const void **)(keys),
                                                              (const void **)(values),
                                                              2,
                                                              &kCFTypeDictionaryKeyCallBacks,
                                                              &kCFTypeDictionaryValueCallBacks);

        CFArrayAppendValue(inputDeviceArray, deviceDictionary);

        CFRelease(deviceDictionary);
        deviceDictionary = NULL;
    }

    free(audioDevices);
    audioDevices = NULL;

    // Return a non-mutable copy of the array
    CFArrayRef immutableInputDeviceArray = CFArrayCreateCopy(kCFAllocatorDefault, inputDeviceArray);
    CFRelease(inputDeviceArray);
    inputDeviceArray = NULL;

    return immutableInputDeviceArray;
}

例如,内置麦克风的 UID 为 BuiltInMicrophoneDevice

iOS 上的音频设备 ID

iOS 不支持通过语音 SDK 来选择音频设备。 使用 SDK 的应用可以通过 AVAudioSession 框架影响音频路由。

例如,可以按照说明

[[AVAudioSession sharedInstance] setCategory:AVAudioSessionCategoryRecord
    withOptions:AVAudioSessionCategoryOptionAllowBluetooth error:NULL];

将蓝牙耳机用于支持语音的应用。

JavaScript 中的音频设备 ID

在 JavaScript 中,MediaDevices.enumerateDevices() 方法可用于枚举媒体设备并查找要传递给 fromMicrophone(...) 的设备 ID。

后续步骤