Pipewire webrtc 回声消除不起作用

Question

我正在编写一个pipewire节点（pipewire版本1.0.0）来同时播放声音并从麦克风录制，音频规格为8kHz采样率，1通道和

SPA_AUDIO_FORMAT_S16

格式。该应用程序可以运行并且还可以播放/录制声音。问题是我启用

libpipewire-module-echo-cancel

并将其配置为使用 webrtc 消除声学回声，因为听不到或记录不到声音。我还更改了应用程序以连接

Echo Cancellation Sink

和

Echo Cancellation Source

节点。管道图似乎没问题：

pw-top的输出：（随着时间的推移

Echo Cancellation Source

和

Echo Cancellation Playback

有误差）

S   ID  QUANT   RATE    WAIT    BUSY   W/Q   B/Q  ERR FORMAT           NAME
S   28      0      0    ---     ---   ---   ---     0                  Dummy-Driver
S   29      0      0    ---     ---   ---   ---     0                  Freewheel-Driver
R   48   2048  48000   6.5ms  10.5us  0.15  0.00    0    S32LE 2 48000 alsa_input.platform-sound.stereo-fallback
R   32   1680   8000 178.4us 136.3us  0.00  0.00    0     F32P 2 48000  + Echo Cancellation Capture
R   33   1680   8000 339.5us  27.0us  0.01  0.00  689     F32P 2 48000  + Echo Cancellation Source
R   34   1680   8000 134.6us 128.2us  0.00  0.00    0     F32P 2 48000  + Echo Cancellation Sink
R   35   1680   8000 429.4us  19.9us  0.01  0.00  689     F32P 2 48000  + Echo Cancellation Playback
R   47      0      0  43.0us   1.1ms  0.00  0.02    0    S32LE 2 48000  + alsa_output.platform-sound.stereo-fallback
R   67      0      0   1.3ms   1.2ms  0.03  0.03    1     S16LE 1 8000  + myapp
R   68      0      0   3.2ms   2.8ms  0.08  0.07    1     S16LE 1 8000  + myapp

pw-link -l :

Echo Cancellation Capture:input_FL
  |<- alsa_input.platform-sound.stereo-fallback:capture_FL
Echo Cancellation Capture:input_FR
  |<- alsa_input.platform-sound.stereo-fallback:capture_FR
Echo Cancellation Source:capture_FL
  |-> myapp:input_FL
Echo Cancellation Source:capture_FR
  |-> myapp:input_FR
Echo Cancellation Sink:playback_FL
  |<- myapp:output_FL
Echo Cancellation Sink:playback_FR
  |<- myapp:output_FR
Echo Cancellation Playback:output_FL
  |-> alsa_output.platform-sound.stereo-fallback:playback_FL
Echo Cancellation Playback:output_FR
  |-> alsa_output.platform-sound.stereo-fallback:playback_FR
alsa_output.platform-sound.stereo-fallback:playback_FL
  |<- Echo Cancellation Playback:output_FL
alsa_output.platform-sound.stereo-fallback:playback_FR
  |<- Echo Cancellation Playback:output_FR
alsa_input.platform-sound.stereo-fallback:capture_FL
  |-> Echo Cancellation Capture:input_FL
alsa_input.platform-sound.stereo-fallback:capture_FR
  |-> Echo Cancellation Capture:input_FR
myapp:input_FL
  |<- Echo Cancellation Source:capture_FL
myapp:input_FR
  |<- Echo Cancellation Source:capture_FR
myapp:output_FL
  |-> Echo Cancellation Sink:playback_FL
myapp:output_FR
  |-> Echo Cancellation Sink:playback_FR

pipewire相关配置：

{   name = libpipewire-module-echo-cancel
      args = {
          library.name  = aec/libspa-aec-webrtc
          node.latency = 1700/8000
          # monitor.mode = false
          capture.props = {
             node.name = "Echo Cancellation Capture"
             #audio.channels = 1
             node.passive = true
             target.object = "alsa_input.platform-sound.stereo-fallback"
          }
          source.props = {
             node.name = "Echo Cancellation Source"
             #audio.channels = 1
          }
          sink.props = {
             node.name = "Echo Cancellation Sink"
             #audio.channels = 1
          }
          playback.props = {
             node.name = "Echo Cancellation Playback"
             #audio.channels = 1
             node.passive = true
             target.object = "alsa_output.platform-sound.stereo-fallback"
          }
      }
    }

我怀疑 8kHz 采样率，因为 aec3 似乎无法使用它，所以改为使用 aecm (

aec.args = {mobile = 1}

) 但仍然无法工作。

另一个问题是，为什么通道数在所有图表和命令输出中显示为 2，而声音是单通道，并且应用程序被编写为与单通道一起使用，如您所见：

params_play[0] = spa_format_audio_raw_build(&b_play, SPA_PARAM_EnumFormat,
            &SPA_AUDIO_INFO_RAW_INIT(
                .format = SPA_AUDIO_FORMAT_S16,
                .channels = 1,
                .rate = 8000 ));

我提醒一下，该应用程序在启用回声消除之前可以工作，但启用后，就听不到或记录不到任何声音。谁能帮我解决这个问题？谢谢。

注意：我用 Yocto 为嵌入式板构建了管道线。

Answer 1

我在 PipeWire 中的回声消除模块中遇到了类似的问题。您是否找到了使其发挥作用的解决方案？