Transfer function measurements for simulating environmental noise at hearable microphones

<p>This dataset is supplementary material to the conference paper "Multi-Microphone Noise Data Augmentation for DNN-based Own Voice Reconstruction for Hearables in Noisy Environments" presented at ICASSP 2024 [1].</p> <p>The dataset consists of impulse response measurements for 18 device users (5 female, 13 male) wearing hearable devices in both ears.&nbsp;<br>The dataset was recorded in a sound-proof listening room using the Hearpiece prototype device (closed vent variant) [2] with a sampling frequency of 44.1 kHz.<br>Impulse responses were measured with exponential sweeps from 80 Hz to 22.05 kHz with a duration of 3s played from 8 loudspeakers arranged in a circle of approximately 1.5m radius.&nbsp;<br>The loudspeakers were located in the horizontal plane around the device users in 45&deg;-steps (azimuth), starting from 22.5&deg; to the right (where 0&deg; is the front from the device users' perspective).</p> <p>The measurements are contained in the folder <code>measurements</code>. Each subfolder contains measurements from a different device user (e.g., <code>VP_01</code>).&nbsp;<br>Each file contains the measurement for one direction, e.g. <code>VP_01/data_0.npz</code> contains the measurement of device user <code>VP_01</code> for 22.5&deg; azimuth, <code>VP_01/data_1.npz</code> is the measurement for the same device user for 22.5&deg;+45&deg; and so on.<br>Measurements of device users where the device could not be inserted, or where the fit did not provide sufficient attenuation of external sounds to the in-ear microphone, were excluded.</p> <p>The impulse responses for two Hearpiece devices (closed vent), the concha and in-ear microphones were measured.<br>A DPA 6060 lavalier clip microphone and a Tbone SC140 cardiod microphone were also included in the measurement as reference channels.&nbsp;</p> <p>The channels of the measurements (counting from 0):</p> <p>&nbsp; &nbsp; 0: Lavalier-microphone clipped to the shirt neck, shirt collar etc. of the device user<br>&nbsp; &nbsp; 1: Reference microphone about 50 cm in front of the device user<br>&nbsp; &nbsp; 2: Left in-ear microphone Hearpiece<br>&nbsp; &nbsp; 3: Left concha microphone Hearpiece<br>&nbsp; &nbsp; 4: Right in-ear microphone Hearpiece<br>&nbsp; &nbsp; 5: Right concha microphone Hearpiece</p> <p><br>The measurement consists of impulse responses from the loudspeaker to the hearable device microphones and reference microphones, and corresponding transfer functions.&nbsp;<br>Measurement metadata is included as well.&nbsp;<br>The measurement files contain a python dictionary with the following fields:</p> <ul> <li><code>test_signal</code>: the signal used for playback, consisting of a pause, the sweep, and another pause</li> <li><code>rec_signal</code>: the recorded signal (sweep played from the loudspeaker, recorded at the microphones)</li> <li><code>sweep</code>: the generated exponential sweep signal without pauses</li> <li><code>T</code>: actual duration of the sweep (~2 Seconds)</li> <li><code>sweep_inv</code>: inverse sweep (inverse w.r.t convolution of the sweep with the system response)</li> <li><code>sweep_inv_spectrum</code>: spectrum of the inverse sweep</li> <li><code>f11</code>: the frequency (in Hz) corresponding to the <code>RampLen</code> of the fade-in at the beginning of the sweep</li> <li><code>T_desd</code>: desired duration of the sweep in seconds (2 Seconds)</li> <li><code>T_rec</code>: recording duration in seconds (3 Seconds)</li> <li><code>start_frequency</code>: Minimum frequency in the measurement / first frequency in the sweep (80 Hz)</li> <li><code>RampLen</code>: Length of the fade-in ramp applied to the beginning of the sweep (based on a Hanning window) (2048 Samples)</li> <li><code>pre_pause_len</code>: pause time between starting the measurement and sweep playback (88200 Samples)</li> <li><code>after_pause_len</code>: pause time after sweep playback (44100 Samples)</li> <li><code>n_repetitions</code>: Number of repetitions for the measurement (1)</li> <li><code>n_channels</code>: Number of recorded channels including loopback (7 = 4 Hearpiece, 2 reference, 1 loopback)</li> <li><code>coh_mat</code>: Mean Squared Coherence per channel (between the measured sweep and the playback sweep signal), has shape (frequencies up to <code>samplerate</code>/2 x channels)</li> <li><code>ir_loopback</code>: the measured impulse response of the loopback channel, used to measure and compensate system delay from audio interface</li> <li><code>ir_mic</code>: the measured impulse responses of the hearable and reference microphones, with shape (samples, channels)</li> <li><code>tf_mic</code>: the measured transfer functions between the loudspeaker and the hearable and reference microphones, with shape (frequencies up to <code>samplerate</code>/2, channels)</li> <li><code>system_delay</code>: the measured system delay from the audio interface (position of the peak of the correlation between playback sweep and loopback sweep signals)</li> <li><code>samplerate</code>: The sampling rate used for the measurements (44100 Hz)</li> </ul> <p>This dataset is compatible with the German own voice recordings available at <a href="../records/10844599" target="_blank" rel="noopener">https://zenodo.org/records/10844599</a> (same participants+device insertion and measurement setup).</p> <p>The example script <code>generate_indiv_noise_dataset.py</code> can be used to augment a single-channel noise dataset to obtain simulated individual hearable noise signals,<br>similar to [1] but using impulse responses directly as filters instead of first computing relative transfer functions and then applying them in the STFT domain.</p> <p><br>[1] M. Ohlenbusch, C. Rollwage, S. Doclo: "Multi-microphone Noise Data Augmentation for DNN-based Own Voice Reconstruction for Hearables in Noisy Environments". In: Proc. IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP). Seoul, South Korea, Apr. 2024, pp. 416-420.<br>[2] F. Denk, M. Lettau, H. Schepker, S. Doclo, R. Roden, M. Blau, J.-H. Bach, J. Wellmann, and B. Kollmeier: "A One-Size-Fits-All Earpiece with Multiple Microphones and Drivers for Hearing Device Research". In: Proc. AES International Conference on Headphone Technology. San Francisco, USA, Aug. 2019.</p>