使用 Superpowered 或 AudioKit 音频库在 Swift 中录制回调函数
Record callback function in Swift with Superpowered or AudioKit audio libraries
我的应用程序(在 swift 中编码)根据音频信号进行实时处理。
我需要一个函数,其中左右缓冲区来自输入(USB 麦克风的 2 个通道),一个函数带有输出缓冲区(也有 2 个通道)。
我曾经使用过 EZAudio,但我在使用 2 通道 96K 格式时遇到内存问题。当 EZAudio 停止时,我想更改为 Superpowered 或 Audiokit。
我的问题是:我无法在这些库中的任何一个中获取带有缓冲区的函数。
超能力:
我在桥头中添加了#import "SuperpoweredIOSAudioIO.h"。
我在 ViewController 中添加了 SuperpoweredIOSAudioIODelegate。
这会自动添加中断、权限和 mapchannels 函数,但不会添加 audioProcessingCallback。
我尝试了以下操作:
audio = SuperpoweredIOSAudioIO(delegate: self, preferredBufferSize: 12, preferredMinimumSamplerate: 96000, audioSessionCategory: AVAudioSessionCategoryPlayAndRecord, channels: 2, audioProcessingCallback: audioProcessingCallback, clientdata: UnsafeMutablePointer)
audio.start()
和
func audioProcessingCallback(buffers: UnsafeMutablePointer<UnsafeMutablePointer<Float>>, inputChannels: UInt32, outputChannels: UInt32, numberOfSamples: UInt32, sampleRate: UInt32, hostTime: UInt64) -> Bool {
return true
}
但我收到错误:
Cannot convert value of type
'(UnsafeMutablePointer>, UInt32, UInt32,
UInt32, UInt32, UInt64) -> Bool' to expected argument type
'audioProcessingCallback!' (aka
'ImplicitlyUnwrappedOptional<@convention(c)
(Optional,
Optional>>>,
UInt32, UInt32, UInt32, UInt32, UInt64) -> Bool>')
我找不到这个库的任何示例 Swift...
使用 AudioKit,这是我所做的:
let mic = AKMicrophone()
installTap(mic)
AudioKit.output = mic
AudioKit.start()
func installTap(_ input:AKNode) {
input.avAudioNode.installTap(onBus: 0, bufferSize: 1024, format: AudioKit.format) { [weak self] (buffer, time) -> Void in
self?.signalTracker(didReceivedBuffer: buffer, atTime: time)
}
}
func signalTracker(didReceivedBuffer buffer: AVAudioPCMBuffer, atTime time: AVAudioTime){
let samples = UnsafeBufferPointer(start: buffer.floatChannelData?[0], count:1024)
audioProcess.ProcessDataCaptureWithBuffer(samples, numberOfSamples: UInt32(1024))
}
它可以在我的算法中获取即将到来的缓冲区,但它似乎不在 "realtime" 中,我的意思是,非常慢..(抱歉,很难解释。)
谢谢!
如果您需要进行实时处理,则不应使用 Swift(或 ObjC)。目前在 AudioKit 中执行此操作的方法是创建一个 AUAudioUnit 子类并在其中进行处理。但是,如果您只需要更快的音频分路器,那么 AKLazyTap 是一个很好的解决方案。它与普通的 tap 不同,因为您必须轮询它以获取数据,但此方法允许重复使用缓冲区,因此您可以根据需要尽快调用它。
下面是使用 AKLazyTap 获取峰值的示例:
import UIKit
import AudioKit
class ViewController: UIViewController {
let microphone = AKMicrophone()
var tap: AKLazyTap?
override func viewDidLoad() {
super.viewDidLoad()
AudioKit.output = microphone
AKSettings.ioBufferDuration = 0.002 // This is to decrease latency for faster callbacks.
tap = AKLazyTap(node: microphone.avAudioNode)
guard tap != nil,
let buffer = AVAudioPCMBuffer(pcmFormat: microphone.avAudioNode.outputFormat(forBus: 0), frameCapacity: 44100) else {
fatalError()
}
// Your timer should fire equal to or faster than your buffer duration
Timer.scheduledTimer(withTimeInterval: AKSettings.ioBufferDuration / 2, repeats: true) { _ in
var timeStamp = AudioTimeStamp()
self.tap?.fillNextBuffer(buffer, timeStamp: &timeStamp)
if buffer.frameLength == 0 { return } // This is important, since we're polling for samples, sometimes it's empty, and sometimes it will be double what it was the last call.
let leftMono = UnsafeBufferPointer(start: buffer.floatChannelData?[0], count:Int(buffer.frameLength))
var peak = Float(0)
for sample in leftMono {
peak = max(peak, fabsf(sample))
}
print("number of samples \(buffer.frameLength) peak \(peak)")
}
AudioKit.start()
}
}
Superpowered 是 C++ API,因为 Swift 不推荐用于实时处理。用 C++ 编写音频处理代码。 "Taps" 很慢,使用 SuperpoweredIOSAudioIO 实时获取音频。
我的应用程序(在 swift 中编码)根据音频信号进行实时处理。
我需要一个函数,其中左右缓冲区来自输入(USB 麦克风的 2 个通道),一个函数带有输出缓冲区(也有 2 个通道)。
我曾经使用过 EZAudio,但我在使用 2 通道 96K 格式时遇到内存问题。当 EZAudio 停止时,我想更改为 Superpowered 或 Audiokit。
我的问题是:我无法在这些库中的任何一个中获取带有缓冲区的函数。
超能力: 我在桥头中添加了#import "SuperpoweredIOSAudioIO.h"。
我在 ViewController 中添加了 SuperpoweredIOSAudioIODelegate。 这会自动添加中断、权限和 mapchannels 函数,但不会添加 audioProcessingCallback。
我尝试了以下操作:
audio = SuperpoweredIOSAudioIO(delegate: self, preferredBufferSize: 12, preferredMinimumSamplerate: 96000, audioSessionCategory: AVAudioSessionCategoryPlayAndRecord, channels: 2, audioProcessingCallback: audioProcessingCallback, clientdata: UnsafeMutablePointer)
audio.start()
和
func audioProcessingCallback(buffers: UnsafeMutablePointer<UnsafeMutablePointer<Float>>, inputChannels: UInt32, outputChannels: UInt32, numberOfSamples: UInt32, sampleRate: UInt32, hostTime: UInt64) -> Bool {
return true
}
但我收到错误:
Cannot convert value of type '(UnsafeMutablePointer>, UInt32, UInt32, UInt32, UInt32, UInt64) -> Bool' to expected argument type 'audioProcessingCallback!' (aka 'ImplicitlyUnwrappedOptional<@convention(c) (Optional, Optional>>>, UInt32, UInt32, UInt32, UInt32, UInt64) -> Bool>')
我找不到这个库的任何示例 Swift...
使用 AudioKit,这是我所做的:
let mic = AKMicrophone()
installTap(mic)
AudioKit.output = mic
AudioKit.start()
func installTap(_ input:AKNode) {
input.avAudioNode.installTap(onBus: 0, bufferSize: 1024, format: AudioKit.format) { [weak self] (buffer, time) -> Void in
self?.signalTracker(didReceivedBuffer: buffer, atTime: time)
}
}
func signalTracker(didReceivedBuffer buffer: AVAudioPCMBuffer, atTime time: AVAudioTime){
let samples = UnsafeBufferPointer(start: buffer.floatChannelData?[0], count:1024)
audioProcess.ProcessDataCaptureWithBuffer(samples, numberOfSamples: UInt32(1024))
}
它可以在我的算法中获取即将到来的缓冲区,但它似乎不在 "realtime" 中,我的意思是,非常慢..(抱歉,很难解释。)
谢谢!
如果您需要进行实时处理,则不应使用 Swift(或 ObjC)。目前在 AudioKit 中执行此操作的方法是创建一个 AUAudioUnit 子类并在其中进行处理。但是,如果您只需要更快的音频分路器,那么 AKLazyTap 是一个很好的解决方案。它与普通的 tap 不同,因为您必须轮询它以获取数据,但此方法允许重复使用缓冲区,因此您可以根据需要尽快调用它。
下面是使用 AKLazyTap 获取峰值的示例:
import UIKit
import AudioKit
class ViewController: UIViewController {
let microphone = AKMicrophone()
var tap: AKLazyTap?
override func viewDidLoad() {
super.viewDidLoad()
AudioKit.output = microphone
AKSettings.ioBufferDuration = 0.002 // This is to decrease latency for faster callbacks.
tap = AKLazyTap(node: microphone.avAudioNode)
guard tap != nil,
let buffer = AVAudioPCMBuffer(pcmFormat: microphone.avAudioNode.outputFormat(forBus: 0), frameCapacity: 44100) else {
fatalError()
}
// Your timer should fire equal to or faster than your buffer duration
Timer.scheduledTimer(withTimeInterval: AKSettings.ioBufferDuration / 2, repeats: true) { _ in
var timeStamp = AudioTimeStamp()
self.tap?.fillNextBuffer(buffer, timeStamp: &timeStamp)
if buffer.frameLength == 0 { return } // This is important, since we're polling for samples, sometimes it's empty, and sometimes it will be double what it was the last call.
let leftMono = UnsafeBufferPointer(start: buffer.floatChannelData?[0], count:Int(buffer.frameLength))
var peak = Float(0)
for sample in leftMono {
peak = max(peak, fabsf(sample))
}
print("number of samples \(buffer.frameLength) peak \(peak)")
}
AudioKit.start()
}
}
Superpowered 是 C++ API,因为 Swift 不推荐用于实时处理。用 C++ 编写音频处理代码。 "Taps" 很慢,使用 SuperpoweredIOSAudioIO 实时获取音频。