Merge pull request #83 from interactive-sonification/develop

Develop

Author: wiccy46

.github/workflows/pya-ci.yaml

@@ -13,10 +13,11 @@ on:
 jobs:
   unit-test:
     runs-on: ${{ matrix.os  }}
+    timeout-minutes: 120
     strategy:
       matrix:
         os: ["ubuntu-latest", "macos-latest"]
-        python-version: ["3.7", "3.8", "3.9", "3.10"]
+        python-version: ["3.8", "3.9", "3.10"]
     steps:
     - uses: actions/checkout@v3
     - name: Install portaudio Ubuntu

Changelog.md

@@ -1,5 +1,11 @@
 # Changelog
 
+## 0.5.2 (Nov 2023)
+* #82, `pyaudio` is now optional: If you plan to use `PyAudioBackend`, install `pya` with `pip install pya[pyaudio]`.
+* Fix audio device bug
+* #77, Improve code type hint
+* #79, Use pyamapping 
+
 ## 0.5.1 (Dec 2022)
 * Now support Python3.10
 * Bugfix #67: When the channels argument of Aserver and Arecorder has not been set it was determined by the default device instead of the actual device.

README.md

@@ -47,7 +47,15 @@ At this time pya is more suitable for offline rendering than realtime.
 
 ## Installation
 
-`pya` requires `portaudio` and its Python wrapper `PyAudio` to play and record audio. 
+Install using
+```
+pip install pya
+```
+
+However to play and record audio you need a backend.
+
+-  `pip install pya[remote]` for a web based Jupyter backend
+-  `pip install pya[pyaudio]` for `portaudio` and its Python wrapper `PyAudio`
 
 ### Using Conda
 
@@ -101,7 +109,7 @@ For Apple ARM Chip, if you failed to install the PyAudio dependency, you can fol
 Try `sudo apt-get install portaudio19-dev` or equivalent to your distro, then 
 
 ```
-pip isntall pya
+pip install pya
 ```
 
 ### Using PIP (Windows)

appveyor.yml

@@ -10,7 +10,7 @@ branches:
 
 environment:
   matrix:
-    - PYTHON_VERSION: 3.7
+    - PYTHON_VERSION: 3.10
       MINICONDA: C:\Miniconda3-x64
 
 init:
@@ -24,7 +24,7 @@ install:
   - conda config --append channels conda-forge
   - "conda create -q -n test-environment python=%PYTHON_VERSION% ffmpeg coverage --file=requirements_remote.txt --file=requirements_test.txt"
   - activate test-environment
-  - "pip install -r requirements.txt"
+  - "pip install -r requirements.txt -r requirements_pyaudio.txt"
 
 test_script:
   - pytest

dev/generate_doc

@@ -80,6 +80,7 @@ def generate_tree(out_folder, publish=False, branches=['master', 'develop'], tag
         res = os.system(f'git -C {doctree_root} checkout {target} -f' + SILENCE)
         if res != 0:
             raise RuntimeError(f'Could not checkout {d}. Git returned status code {res}!')
+        os.system(f'cp {template_folder}/../conf.py {doctree_root}/docs/')
         if template_folder:
             os.system(f'cp {template_folder}/* {doctree_root}/docs/_templates')
 

examples/pya-examples.ipynb

@@ -23,7 +23,8 @@
     "    sys.path.insert(0, cmd_folder)\n",
     "\n",
     "# ... the part relevant for pya usage starts here\n",
-    "from pya import *"
+    "from pya import *\n",
+    "from pyamapping import dbamp, ampdb"
    ]
   },
   {
@@ -2123,7 +2124,7 @@
    "name": "python",
    "nbconvert_exporter": "python",
    "pygments_lexer": "ipython3",
-   "version": "3.9.14"
+   "version": "3.10.7"
   },
   "toc": {
    "base_numbering": 1,

pya/amfcc.py

@@ -1,10 +1,14 @@
-import numpy as np
+from typing import Optional, Union
 from warnings import warn
-from .helper import next_pow2, signal_to_frame, round_half_up, magspec
-from .helper import mel2hz, hz2mel, is_pow2
-from .helper import basicplot
+
+import numpy as np
+from pyamapping import mel_to_hz, hz_to_mel
 from scipy.signal import get_window
 from scipy.fftpack import dct
+
+from .helper import next_pow2, signal_to_frame, round_half_up, magspec
+from .helper import is_pow2
+from .helper import basicplot
 import pya.asig
 import logging
 
@@ -64,10 +68,12 @@ class Amfcc:
         An array of the MFCC coeffcient, size: nframes x ncep
     """
 
-    def __init__(self, x, sr=None, label='', n_per_frame=None,
-                 hopsize=None, nfft=None, window='hann', nfilters=26,
-                 ncep=13, ceplifter=22, preemph=0.95,
-                 append_energy=True, cn=None):
+    def __init__(self, x: Union[pya.Asig, np.ndarray], sr: Optional[int] = None,
+                 label: str = '', n_per_frame: Optional[int] = None,
+                 hopsize: Optional[int] = None, nfft: Optional[int] = None,
+                 window: str = 'hann', nfilters: int = 26,
+                 ncep: int = 13, ceplifter: int = 22, preemph: float = 0.95,
+                 append_energy: bool = True, cn: Optional[list] = None):
         """Initialize Amfcc object
 
         Parameters
@@ -103,14 +109,14 @@ def __init__(self, x, sr=None, label='', n_per_frame=None,
         append_energy : bool
             If true, the zeroth cepstral coefficient is replaced with the log
             of the total frame energy.
-        cn : list
-            A list of channel name based on the Asig.
+        cn : list or None
+            A list of channel names, size should match the channels.
         """
         # ----------Prepare attributes ------------`-------------
         # First prepare for parameters
         # x represent the audio signal, which can be Asig object or np.array.
         self.im = None
-        if type(x) == pya.asig.Asig:
+        if isinstance(x, pya.asig.Asig):
             self.sr = x.sr
             self.x = x.sig
             self.label = ''.join([x.label, "_mfccs"])
@@ -129,9 +135,9 @@ def __init__(self, x, sr=None, label='', n_per_frame=None,
             self.duration = np.shape(x)[0] / self.sr
             self.label = label
             self.channels = 1 if self.x.ndim == 1 else self.x.shape[1]
-            self.cn = None
+            self.cn = cn
         else:
-            msg = "x can only be either a numpy.ndarray or pya.Asig object."
+            msg = "x can only be either a numpy.ndarray or Asig object."
             raise TypeError(msg)
 
         # default 25ms length window.
@@ -211,7 +217,7 @@ def __repr__(self):
         return f"Amfcc({self.label}): sr {self.sr}, length: {self.duration} s"
 
     @staticmethod
-    def preemphasis(x, coeff=0.97):
+    def preemphasis(x: np.ndarray, coeff: float = 0.97):
         """Pre-emphasis filter to whiten the spectrum.
         Pre-emphasis is a way of compensating for the
         rapid decaying spectrum of speech.
@@ -233,7 +239,8 @@ def preemphasis(x, coeff=0.97):
         return np.append(x[0], x[1:] - coeff * x[:-1])
 
     @staticmethod
-    def mel_filterbanks(sr, nfilters=26, nfft=512, lowfreq=0, highfreq=None):
+    def mel_filterbanks(sr: int, nfilters: int = 26, nfft: int = 512,
+                        lowfreq: float = 0, highfreq: Optional[float] = None):
         """Compute a Mel-filterbank. The filters are stored in the rows,
         the columns correspond to fft bins. The filters are returned as
         an array of size nfilt * (nfft/2 + 1)
@@ -246,9 +253,9 @@ def mel_filterbanks(sr, nfilters=26, nfft=512, lowfreq=0, highfreq=None):
             The number of filters, default 20
         nfft : int
             The size of FFT, default 512
-        lowfreq : int or float
+        lowfreq : float
             The lowest band edge of the mel filters, default 0 Hz
-        highfreq : int or float
+        highfreq : float
             The highest band edge of the mel filters, default sr // 2
 
         Returns
@@ -260,12 +267,12 @@ def mel_filterbanks(sr, nfilters=26, nfft=512, lowfreq=0, highfreq=None):
         highfreq = highfreq or sr // 2
 
         # compute points evenly spaced in mels
-        lowmel = hz2mel(lowfreq)
-        highmel = hz2mel(highfreq)
+        lowmel = hz_to_mel(lowfreq)
+        highmel = hz_to_mel(highfreq)
         melpoints = np.linspace(lowmel, highmel, nfilters + 2)
         # our points are in Hz, but we use fft bins, so we have to convert
         #  from Hz to fft bin number
-        bin = np.floor((nfft + 1) * mel2hz(melpoints) / sr)
+        bin = np.floor((nfft + 1) * mel_to_hz(melpoints) / sr)
 
         filter_banks = np.zeros([nfilters, nfft // 2 + 1])
         for j in range(0, nfilters):
@@ -276,7 +283,7 @@ def mel_filterbanks(sr, nfilters=26, nfft=512, lowfreq=0, highfreq=None):
         return filter_banks
 
     @staticmethod
-    def lifter(cepstra, L=22):
+    def lifter(cepstra: np.ndarray, L: int = 22):
         """Apply a cepstral lifter the the matrix of cepstra.
         This has the effect of increasing the magnitude of
         the high frequency DCT coeffs.
@@ -315,21 +322,23 @@ def lifter(cepstra, L=22):
             # values of L <= 0, do nothing
             return cepstra
 
-    def plot(self, cmap='inferno', show_bar=True,
-             offset=0, scale=1., xlim=None, ylim=None,
-             x_as_time=True, nxlabel=8, ax=None, **kwargs):
+    def plot(self, show_bar: bool = True, offset: int = 0, scale: float = 1.,
+             xlim: Optional[float] = None, ylim: Optional[float] = None,
+             x_as_time: bool = True, nxlabel: int = 8, ax=None, **kwargs):
         """Plot Amfcc.features via matshow, x is frames/time, y is the MFCCs
 
         Parameters
         ----------
-        figsize : (float, float), optional, default: None
-            Figure size, width, height in inches, Default = [6.4, 4.8]
-        cmap : str
-            colormap for matplotlib. Default is 'inferno'.
         show_bar : bool, optional
             Default is True, show colorbar.
-        x_as_time : bool, optional
-            Default is True, show x axis as time or sample index.
+        offset: int
+            It is the spacing between channel, without setting it every channel will be overlayed onto each other.
+        scale: float
+            Visual scaling for improve visibility
+        xlim: float, optional
+            x axis value range limit
+        ylim: float, optional
+            y axis value range limit
         nxlabel : int, optional
             The amountt of labels on the x axis. Default is 8 .
         """

pya/arecorder.py

@@ -1,13 +1,13 @@
 # Arecorder class
-import time
 import logging
 import numbers
-from warnings import warn
+import time
+from typing import Optional, Union
+
 import numpy as np
-import pyaudio
 from . import Asig
 from . import Aserver
-from .helper import dbamp
+from pyamapping import db_to_amp
 
 
 _LOGGER = logging.getLogger(__name__)
@@ -30,19 +30,33 @@ class Arecorder(Aserver):
     [Asig(''): ... x ... @ 44100Hz = ...
     """
 
-    def __init__(self, sr=44100, bs=256, device=None, channels=None, backend=None, **kwargs):
+    def __init__(self, sr: int = 44100, bs: int = 256, device: Optional[int] = None,
+                 channels: Optional[int] = None, backend=None, **kwargs):
         super().__init__(sr=sr, bs=bs, device=device, 
                          backend=backend, **kwargs)
         self.record_buffer = []
         self.recordings = []  # store recorded Asigs, time stamp in label
         self._recording = False
         self._record_all = True
-        self.gains = np.ones(self.channels)
         self.tracks = slice(None)
-        self._device = device or self.backend.get_default_input_device_info()['index']
-        self.channels = channels or self.backend.get_device_info_by_index(self._device)['maxInputChannels']
+        self._device = self.backend.get_default_input_device_info()['index'] if device is None else device
+        self._channels = channels or self.max_in_chn
+        self.gains = np.ones(self._channels)
+
+    @property
+    def channels(self):
+        return self._channels
+
+    @channels.setter
+    def channels(self, val: int):
+        """
+        Set the number of channels. Aserver needs reboot.
+        """
+        if val > self.max_in_chn:
+            raise ValueError(f"AServer: channels {val} > max {self.max_in_chn}")
+        self._channels = val
 
-    def set_tracks(self, tracks, gains):
+    def set_tracks(self, tracks: Union[list, np.ndarray], gains: Union[list, np.ndarray]):
         """Define the number of track to be recorded and their gains.
 
         parameters
@@ -58,10 +72,10 @@ def set_tracks(self, tracks, gains):
             elif len(tracks) > self.channels or len(gains) > self.channels:
                 raise AttributeError("argument cannot be larger than channels.")
             self.tracks = tracks
-            self.gains = np.array([dbamp(g) for g in gains], dtype="float32")
+            self.gains = np.array([db_to_amp(g) for g in gains], dtype="float32")
         elif isinstance(tracks, numbers.Number) and isinstance(gains, numbers.Number):
             self.tracks = [tracks]
-            self.gains = dbamp(gains)
+            self.gains = db_to_amp(gains)
         else:
             raise TypeError("Arguments need to be both list or both number.")
 
@@ -70,8 +84,6 @@ def reset(self):
         self.gains = np.ones(self.channels)
 
     def boot(self):
-        """boot recorder"""
-        # when input = True, the channels refers to the input channels.
         self.boot_time = time.time()
         self.block_time = self.boot_time
         # self.block_cnt = 0
@@ -95,7 +107,7 @@ def _recorder_callback(self, in_data, frame_count, time_info, flag):
             self.record_buffer.append(data_float)
             # E = 10 * np.log10(np.mean(data_float ** 2)) # energy in dB
             # os.write(1, f"\r{E}    | {self.block_cnt}".encode())
-        return None, pyaudio.paContinue
+        return self.backend.process_buffer(None)
 
     def record(self):
         """Activate recording"""