Generating wav about pyroomacoustics

[fakufaku]

Hi @maelp , I'll answer in order.

1. That's a perfectly valid thing to want! I think it should be possible to do this without too much effort. I will try to write an example in the next few days.
2. Wether one wants to use 2D or 3D is entirely dependent on the situation I guess. The situation in research in acoustical signal processing is that many people validate their algorithms in 2D room environments. Now I think this has more to do with ease of implementation than accuracy. You can imagine that a 2D room is like a room without floor or ceiling (an infinite elevator shaft). If you are interested in realistic rendering, I would use a 3D room. Then again, the image source model is not the most life-like simulator to use.
3. I have not used the simulator from Eric Lehmann, but if I recall correctly, it uses a more sophisticated method to improve the rendering of the tail of the RIR, but has the drawback of only handling shoebox rooms and requires Matlab.

[fakufaku]

Closed by mistake. I'll close the issue when the example is added :)

[maelp]

Thanks Robin On 8 October 2017 at 09:13:29, Robin Scheibler ([email protected]) wrote: Reopened #8 <#8>. — You are receiving this because you were mentioned. Reply to this email directly, view it on GitHub <#8 (comment)>, or mute the thread <https://github.com/notifications/unsubscribe-auth/AAA8Yu4htfZWxwFX2l0tR83BnlSUkL6yks5sqHYYgaJpZM4PvQyq> .

[fakufaku]

Finally, I could do this quickly. I have added room_simulation.py in the examples folder. I hope this helps.

'''
A simple example of using pyroomacoustics to simulate
sound propagation in a shoebox room and record the result
to a wav file.
'''

import numpy as np
import pyroomacoustics as pra
import matplotlib.pyplot as plt
from scipy.io import wavfile

fs, audio_anechoic = wavfile.read('examples/samples/guitar_16k.wav')

# room dimension
room_dim = [5, 4, 6]

# Create the shoebox
shoebox = pra.ShoeBox(
    room_dim,
    absorption=0.2,
    fs=fs,
    max_order=15,
    )

# source and mic locations
shoebox.add_source([2, 3.1, 2], signal=audio_anechoic)
shoebox.add_microphone_array(
        pra.MicrophoneArray(
            np.array([[2, 1.5, 2]]).T, 
            shoebox.fs)
        )

# run ism
shoebox.simulate()

audio_reverb = shoebox.mic_array.to_wav('examples/samples/guitar_16k_reverb.wav', norm=True, bitdepth=np.int16)

[maelp]

Thanks @fakufaku, would this be correct to simulate the reverb:

• create the room
• create a source which emits a single sample with amplitude 1.0
• get the audio_reverb
• convolve wav with it

[fakufaku]

Dear @maelp , as I understand, you are interested to look at the room impulse response of the room. There is a simpler way to do this. When you run the shoebox.simulate() line above, all room impulse responses will be computed and stored in shoebox.rir. Then, the impulse response between the m-th microphone and the s-th source can be accessed at shoebox.rir[m][s].

In case you are only interested in getting the impulse response (as opposed to simulate the propagation of a particular sound sample), you can omit the source signal specification and call shoebox.compute_rir() instead of simulate(). Here is the modified example.

'''
A simple example of using pyroomacoustics to compute
the room impulse response between one source and one microphone.
'''

import numpy as np
import pyroomacoustics as pra
import matplotlib.pyplot as plt

# room dimension
room_dim = [5, 4, 6]

# Create the shoebox
shoebox = pra.ShoeBox(
    room_dim,
    absorption=0.2,
    fs=fs,
    max_order=15,
    )

# source and mic locations
shoebox.add_source([2, 3.1, 2])
shoebox.add_microphone_array(
        pra.MicrophoneArray(
            np.array([[2, 1.5, 2]]).T, 
            shoebox.fs)
        )

# run ism
shoebox.compute_rir()

# this plots the RIR between the 1st source and the 1st microphone
rir_time = np.arange(len(shoebox.rir[0][0])) / shoebox.fs
plt.plot(rir_time, shoebox.rir[0][0])

[maelp]

Thanks !