VoiceClean

The main pipeline class that chains AEC and VAD together.

from voiceclean import VoiceClean

VoiceClean

class VoiceClean(
    sample_rate: int = 8000,
    vad_threshold: float = 0.5,
)

Full audio cleanup pipeline: AEC followed by VAD. Components are optional — if onnxruntime is not installed, VAD is skipped with a warning.

Parameters:

Parameter	Type	Default	Description
sample_rate	int	8000	Audio sample rate in Hz
vad_threshold	float	0.5	Speech probability threshold for VAD

Properties

has_aec

@property
def has_aec(self) -> bool

Returns True if AEC is available (always True since AEC has no optional dependencies).

has_vad

@property
def has_vad(self) -> bool

Returns True if VAD is available (onnxruntime is installed).

Methods

feed_reference

def feed_reference(self, audio: bytes) -> None

Feed the bot's outgoing audio as AEC reference signal. Call this every time the bot sends audio to the speaker.

Parameters:

• audio — Raw PCM int16 mono bytes at the configured sample rate.

process

def process(self, mic_audio: bytes) -> ProcessResult

Run the full pipeline (AEC then VAD) on mic audio.

Parameters:

• mic_audio — Raw PCM int16 mono bytes from the caller's mic.

Returns: ProcessResult

ProcessResult

@dataclass
class ProcessResult:
    audio: bytes
    is_speech: bool
    speech_prob: float

Field	Type	Description
audio	bytes	Cleaned PCM int16 mono audio (echo removed)
is_speech	bool	Whether the caller is speaking
speech_prob	float	Speech probability from 0.0 to 1.0

Example

from voiceclean import VoiceClean

vc = VoiceClean(sample_rate=8000)

# Audio processing loop
while True:
    bot_audio = get_bot_output()    # TTS audio being sent to caller
    mic_audio = get_mic_input()     # audio from caller's mic

    vc.feed_reference(bot_audio)
    result = vc.process(mic_audio)

    if result.is_speech:
        send_to_stt(result.audio)   # only transcribe when caller is speaking