[Python] Pyaudio 教學

程式語言：Python
Package：pyaudio
官方網站
官方文件

功能：音訊的播放與錄製

基本知識

sampling rate

類比訊號轉成數位訊號的取樣率

bit depth

用多少 bit 表現一個 sample

channel

聲道
1 => mono 為單聲道
2 => stereo 為雙聲道或又稱立體聲

frame

構成一個聲音的最小單位
即是一組 samples。與 channel 相關
單聲道 ( mono ) = 1 sample = 1 frame
雙聲道 ( stereo) = 2 samples = 1 frame
5.1聲道(左, 中央, 右, 右後, 左後, 低音) = 6 samples = 1 frame
frame size = bit depth * channels

Blocking 範例

'''
播放特定頻率
'''
import numpy as np
import pyaudio


def sine(frequency, t, sampleRate):
    '''
    產生 sin wave

    :Args:
     - frequency: 欲產生的頻率 Hz
     - t: 播放時間長度 seconds
     - sampleRate: 取樣頻率 1/s
    '''
    # 播放數量
    n = int(t * sampleRate)
    # 每秒轉動的角度再細分為取樣間隔
    interval = 2 * np.pi * frequency / sampleRate
    return np.sin(np.arange(n) * interval)


def play_tone(stream, frequency=440, t=1, sampleRate=44100):
    '''
    播放特定頻率

    :Args:
     - stream: 
     - frequency: 欲產生的頻率 Hz
     - t: 播放時間長度 seconds
     - sampleRate: 取樣頻率 1/s
    '''
    data = sine(frequency, t, sampleRate)
    
    # 因 format 為  pyaudio.paFloat32，故轉換為 np.float32 並轉換為 bytearray
    stream.write(data.astype(np.float32).tostring())


if __name__ == '__main__':
    p = pyaudio.PyAudio()
    stream = p.open(format=pyaudio.paFloat32,
                    channels=1, rate=44100, output=True)

    play_tone(stream, 
     frequency=1000, #Hz
     t=3) #seconds

    stream.close()
    p.terminate()

Non-Blocking 範例

'''
non-blocking
播放特定頻率
'''
import numpy as np
import pyaudio
import time

# frames
CHUNK = 1024
# channels
CH = 1

def sine(frequency, t, sampleRate):
    '''
    產生 sin wave

    :Args:
     - frequency: 欲產生的頻率 Hz
     - t: 播放時間長度 seconds
     - sampleRate: 取樣頻率 1/s
    '''
    # 播放數量
    n = int(t * sampleRate)
    # 每秒轉動的角度再細分為取樣間隔
    interval = 2 * np.pi * frequency / sampleRate
    
    return np.sin(np.arange(n) * interval)

    
def sliceData(frame_count, channels):  
    '''
    切出合適的長度，也就是 frame_count * channels * sampleBytes

    :Args:
     - frame_count: frames 的數目
     - channels: channels 數目
    '''
    data = sine(frequency=1000, t=3, sampleRate=44100)
    
    # 因會再轉換為 np.float32，故無需乘上 sampleBytes
    size = channels * frame_count
    while True:
        dataSlice = data[:size]
        # 此時小數點會用 np.float32 4byte 表示，故資料長度會變為 4 倍
        dataBytes = dataSlice.astype(np.float32).tostring()
        
        yield dataBytes
        # 刪除已輸出資料
        data = np.delete(data, range(size))

    
dataGen = sliceData(CHUNK, CH)
def callback(in_data, frame_count, time_info, status):
    global dataGen
    data = next(dataGen)

    return (data, pyaudio.paContinue)


if __name__ == '__main__':
    p = pyaudio.PyAudio()
    stream = p.open(format=pyaudio.paFloat32,
                    channels=CH, rate=44100, output=True, frames_per_buffer=CHUNK, stream_callback=callback)

    # stream.start_stream()
    
    for i in range(1000):
        print(i)    
        
    stream.stop_stream()
    stream.close()
    p.terminate()

Module pyaudio

Classes

pyaudio.PyAudio

pyaudio.Stream

Host Specific Classes

pyaudio.PaMacCoreStreamInfo

Stream Conversion Convenience Functions

pyaudio.get_sample_size(format)

指定格式的大小 (bytes)
format

即是下面的 Portaudio Sample Formats

pyaudio.get_format_from_width(width, unsigned=True)

指定 sample 大小的格式

除 pyaudio.paInt32 以外的格式

width

byte 單位，可為 1, 2, 3, 4

unsigned

有無負值

PortAudio version

pyaudio.get_portaudio_version()

portaudio 版本 (Int)
"19.5.1" 會得到 16進位 0x00130501

pyaudio.get_portaudio_version_text()

portaudio 版本 (String)

Portaudio Sample Formats

pyaudio.paFloat32 = 1
pyaudio.paInt32 = 2
pyaudio.paInt24 = 4
pyaudio.paInt16 = 8
pyaudio.paInt8 = 16
pyaudio.paUInt8 = 32
pyaudio.paCustomFormat = 65536

PortAudio Host APIs

pyaudio.paInDevelopment

Still in development

pyaudio.paDirectSound

DirectSound (Windows only)

pyaudio.paMME

Multimedia Extension (Windows only)

pyaudio.paASIO

Steinberg Audio Stream Input/Output

pyaudio.paSoundManager

SoundManager (OSX only)

pyaudio.paCoreAudio

CoreAudio (OSX only)

pyaudio.paOSS

Open Sound System (Linux only)

pyaudio.paALSA

Advanced Linux Sound Architecture (Linux only)

pyaudio.paAL

Open Audio Library

pyaudio.paBeOS

BeOS Sound System

pyaudio.paWDMKS

Windows Driver Model (Windows only)

pyaudio.paJACK

JACK Audio Connection Kit

pyaudio.paWASAPI

Windows Vista Audio stack architecture

pyaudio.paNoDevice

Not actually an audio device

PortAudio Error Codes

pyaudio.paNoError

pyaudio.paNotInitialized

pyaudio.paUnanticipatedHostError

pyaudio.paInvalidChannelCount

pyaudio.paInvalidSampleRate

pyaudio.paInvalidDevice

pyaudio.paInvalidFlag

pyaudio.paSampleFormatNotSupported

pyaudio.paBadIODeviceCombination

pyaudio.paInsufficientMemory

pyaudio.paBufferTooBig

pyaudio.paBufferTooSmall

pyaudio.paNullCallback

pyaudio.paBadStreamPtr

pyaudio.paTimedOut

pyaudio.paInternalError

pyaudio.paDeviceUnavailable

pyaudio.paIncompatibleHostApiSpecificStreamInfo

pyaudio.paStreamIsStopped

pyaudio.paStreamIsNotStopped

pyaudio.paInputOverflowed

pyaudio.paOutputUnderflowed

pyaudio.paHostApiNotFound

pyaudio.paInvalidHostApi

pyaudio.paCanNotReadFromACallbackStream

pyaudio.paCanNotWriteToACallbackStream

pyaudio.paCanNotReadFromAnOutputOnlyStream

pyaudio.paCanNotWriteToAnInputOnlyStream

pyaudio.paIncompatibleStreamHostApi

PortAudio Callback Return Codes

pyaudio.paContinue

There is more audio data to come

pyaudio.paComplete

This was the last block of audio data

pyaudio.paAbort

An error ocurred, stop playback/recording

PortAudio Callback Flags

pyaudio.paInputUnderflow

Buffer underflow in input

pyaudio.paInputOverflow

Buffer overflow in input

pyaudio.paOutputUnderflow

Buffer underflow in output

pyaudio.paOutputOverflow

Buffer overflow in output

pyaudio.paPrimingOutput

Just priming, not playing yet

class pyaudio.PyAudio

利用此 class 開關 streams
Stream Management

open(rate, channels, format, input=False, output=False, input_device_index=None, output_device_index=None, frames_per_buffer=1024, start=True, input_host_api_specific_stream_info=None, output_host_api_specific_stream_info=None, stream_callback=None)

rate

取樣率 (Hz)

channels

channels 數目

format

即是上面的 Portaudio Sample Formats

input

是否為 input

output

是否為 output

input_device_index

輸入裝置的 index，None 表預設

output_device_index

輸出裝置的 index，None 表預設

frames_per_buffer

buffer 共有多少 frames

start

stream 是否立即開始

input_host_api_specific_stream_info

使用在 PaMacCoreStreamInfo

output_host_api_specific_stream_info

使用在 PaMacCoreStreamInfo

stream_callback

運用在 non-blocking
Function 格式如下

def callback(in_data, # 若 input=True 則傳進錄製的資料，不然為 None
         frame_count, # number of frames
         time_info,   # dictionary
         status_flags)# PortAudio Callback Flags

    # 不可使用 Stream.read() or Stream.write()

    # out_data 長度應為 frame_count * channels * sampleBytes
    return (out_data, flag)

close(stream)

通常使用 Stream.close()

terminate()

中斷 PortAudio 釋放資源

Host API

get_host_api_count()

Host API 的總數量

get_default_host_api_info()

預設 Host API 的資訊

get_host_api_info_by_type(host_api_type)

指定 Host API 的資訊
host_api_type

即是上面的 PortAudio Host APIs

get_host_api_info_by_index(host_api_index)

指定 Host API 的資訊
host_api_index

第幾個 Host API

get_device_info_by_host_api_device_index(host_api_index, host_api_device_index)

指定 Host API 的指定裝置資訊
host_api_index

第幾個 Host API

host_api_device_index

Host API 的第幾個裝置

Device API

get_device_count()

裝置的總數量

is_format_supported(rate, input_device=None, input_channels=None, input_format=None, output_device=None, output_channels=None, output_format=None)

格式是否支援
rate

取樣率 (Hz)

input_device

輸入裝置的 index

input_channels

輸入 channels 數目

input_format

輸入格式

output_device

輸出裝置的 index

output_channels

輸出 channels 數目

output_format

輸出格式

get_default_input_device_info()

預設輸入裝置的資訊

get_default_output_device_info()

預設輸出裝置的資訊

get_device_info_by_index(device_index)

指定裝置的資訊
device_index

第幾個裝置

Stream Format Conversion

get_sample_size(format)

指定格式的大小 (bytes)
format

即是上面的 Portaudio Sample Formats

get_format_from_width(width, unsigned=True)

指定 sample 大小的格式

除 pyaudio.paInt32 以外的格式

width

byte 單位，可為 1, 2, 3, 4

unsigned

有無負值

class pyaudio.Stream

PortAudio Stream Wrapper
需用 PyAudio.open() 建立
Opening and Closing

__init__(PA_manager, rate, channels, format, input=False, output=False, input_device_index=None, output_device_index=None, frames_per_buffer=1024, start=True, input_host_api_specific_stream_info=None, output_host_api_specific_stream_info=None, stream_callback=None)

必須由 PyAudio.open() 呼叫，可為 input or output 甚至皆可
PA_manager

PyAudio instance

rate

取樣率 (Hz)

channels

channels 數目

format

即是上面的 Portaudio Sample Formats

input

是否為 input

output

是否為 output

input_device_index

輸入裝置的 index，None 表預設

output_device_index

輸出裝置的 index，None 表預設

frames_per_buffer

buffer 共有多少 frames

start

stream 是否立即開始

input_host_api_specific_stream_info

使用在 PaMacCoreStreamInfo

output_host_api_specific_stream_info

使用在 PaMacCoreStreamInfo

stream_callback

運用在 non-blocking
Function 格式如下

def callback(in_data, # 若 input=True 則傳進錄製的資料，不然為 None
         frame_count, # number of frames
         time_info,   # dictionary
         status_flags)# PortAudio Callback Flags

    # 不可使用 Stream.read() or Stream.write()

    # out_data 長度應為 frame_count * channels * sampleBytes
    return (out_data, flag)

close()

關閉 stream

Stream Info

get_input_latency()

input 延遲

get_output_latency()

output 延遲

get_time()

stream 時間

get_cpu_load()

CPU 使用率，只適用於 non-blocking

Stream Management

start_stream()

開始執行 stream

stop_stream()

停止執行 stream

is_active()

stream 是否啟動

is_stopped()

stream 是否停止

Input Output

write(frames, num_frames=None, exception_on_underflow=False)

寫入資料於 stream
不可使用在 non-blocking mode
frames

寫入的資料

num_frames

寫入的 frames 數目，None 表示自動計算

exception_on_underflow

underflow 是否丟出 exception

read(num_frames, exception_on_overflow=True)

從 stream 讀取資料
不可使用在 non-blocking mode
num_frames

讀取的 frames 數目

exception_on_overflow

overflow 是否丟出 exception

get_read_available()

無需等待的可讀 frames 數目

get_write_available()

無需等待的可寫 frames 數目

參考

Android Audio 裡的各種設定和其關係 (一)

子風的知識庫

搜尋此網誌

[Python] Pyaudio 教學

基本知識

Module pyaudio

class pyaudio.PyAudio

class pyaudio.Stream

參考

留言

張貼留言