2

我有以下功能：

private func getPCMBuffer(utterance: AVSpeechUtterance,
                          completion: @escaping (Result<AVAudioPCMBuffer, Failures>) -> Void
) {
    speechSynthesizer.write(utterance) { (buffer: AVAudioBuffer) in
        guard let pcmBuffer = buffer as? AVAudioPCMBuffer else {
            fatalError("unknown buffer type: \(buffer)")
        }

        completion(.success(pcmBuffer))
    }
    completion(.failure(.failed))
}

它返回给我一个 AVAudioPCMBuffer。我已经验证了我传入的话语是正确的。

当我尝试将此 AVAudioPCMBuffer 写入本地 URL 时，问题就出现了，如下所示：

                getPCMBuffer(utterance: speechUtterance) { result in
                    switch result {
                    case .success(let buffer):
                        var settings: [String : Any] = [:]
                        let savePathUrl: URL = URL(fileURLWithPath: NSHomeDirectory() + "/Documents/audioFile.caf")

                        settings[AVFormatIDKey] = kAudioFormatMPEG4AAC
                        settings[AVAudioFileTypeKey] = kAudioFileCAFType
                        settings[AVSampleRateKey] = buffer.format.sampleRate
                        settings[AVNumberOfChannelsKey] = 2
                        settings[AVLinearPCMIsFloatKey] = (buffer.format.commonFormat == .pcmFormatInt32)
                        do {
                            let tempFile = try AVAudioFile(forWriting: savePathUrl, settings: settings, commonFormat: buffer.format.commonFormat, interleaved: buffer.format.isInterleaved)
                            try tempFile.write(from: buffer)
                        } catch {
                            print(error)
                        }
                    case .failure(let failure):
                        print(failure.localizedDescription)
                    }
                }

我遇到以下错误：CABufferList.h:179:BytesConsumed: ASSERTION FAILURE [(nBytes <= buf->mDataByteSize) != 0 is false]：

在我尝试做的那一行：

try tempFile.write(from: buffer)

4

1 回答 1

0

有几件事情需要考虑：

  1. 必须使用正确的文件格式。
  2. 的回调writeUtterance:toBufferCallback:被多次调用，每次都带有合成音频的下一个块。据我所知，它每次最多可以合成 10 毫秒的音频。（无证）
  3. 您必须关闭文件才能刷新它。关闭是通过释放对对象的引用来完成的。（无证）。
@property (readonly, nonatomic) AVSpeechSynthesizer *synth;
@property (strong, nonatomic) AVAudioFile *file;

...

_synth = [[AVSpeechSynthesizer alloc] init];
self.synth.delegate = self;

...

- (void)speechSynthesizer:(AVSpeechSynthesizer *)synthesizer
 didFinishSpeechUtterance:(AVSpeechUtterance *)utterance {
  self.file = nil; // releasing the file will close it
  NSLog(@"Finished");
}


- (void)generateTranscript {
  AVSpeechUtterance *utterance =
      [AVSpeechUtterance speechUtteranceWithString:...];
  utterance.voice = ...;

  [[AVAudioSession sharedInstance] setCategory:AVAudioSessionCategoryPlayback
                                          withOptions:AVAudioSessionCategoryOptionMixWithOthers
                                         error:nil];

  [self.synth writeUtterance:utterance toBufferCallback:^(AVAudioBuffer * _Nonnull buffer) {
    NSAssert([buffer isKindOfClass:AVAudioPCMBuffer.class], @"");
    NSError *error;

    if (!self.file) {
      NSURL *url = [NSURL fileURLWithPath:[NSTemporaryDirectory()
                                           stringByAppendingPathComponent:@"recording.m4a"]];
      NSMutableDictionary *settings = [NSMutableDictionary dictionary];
      settings[AVFormatIDKey] = @(kAudioFormatMPEG4AAC);
      settings[AVEncoderAudioQualityKey] = @(AVAudioQualityMax);
      settings[AVSampleRateKey] = buffer.format.settings[AVSampleRateKey]; // 22050 Hz
      settings[AVNumberOfChannelsKey] = buffer.format.settings[AVNumberOfChannelsKey]; // 1 channel
      self.file = [[AVAudioFile alloc] initForWriting:url settings:settings
                                         commonFormat:buffer.format.commonFormat // AVAudioPCMFormatInt16
                                          interleaved:buffer.format.interleaved // YES
                                                        error:&error];
      if (error) {
        NSLog(@"%@", error);
        return;
      }
    }

    [self.file writeFromBuffer:(AVAudioPCMBuffer *)buffer error:&error];
    if (error) {
      NSLog(@"%@", error);
      return;
    }

    AVAudioFrameCount frameCont = ((AVAudioPCMBuffer *)buffer).frameLength;
    double sampleRate = buffer.format.sampleRate;
    NSLog(@"Written %.4lfs", (double)frameCont / sampleRate);
  }];
}

关于格式，我还建议阅读AVSpeechSynthesisVoice audioFileSettings文档。

于 2022-02-07T00:01:35.830 回答