| Over the years devices have become more complex and we found ourselves needing an increasing number of artificial speech sources to simulate interfering talkers. The IAR Mouth Source was developed as a low-cost, accessible ITU-T P.51 compliant acoustic source for testing. It aims to replicate human speech directivity while being easy to manufacture and maintain. The design balances performance, manufacturability, and compliance with international standards- and while we're currently using it in some measurement setups, there is still a list of improvements desired. |  |
| Goals: Accessible and Standards-Compliant The primary goals were:
COMSOL was used to model the front volume and opening geometry, which significantly affect nearfield acoustic performance |  |
Rapid iteration of geometries resulted in a finalized design that would ensure compliance with P.51.
The mouth prototype was built and compared to limits and simulation at both nearfield and farfield locations.
Calibration at the mouth reference point (MRP) verified frequency response (FR) and total harmonic distortion (THD):
Outcomes: Bandwidth Gains with Trade-Offs
The prototype mouth achieved improved high-frequency bandwidth with no change in low-frequency performance. A limiting factor was the Helmholtz resonance near 2 kHz, caused by the mouth opening interacting with the front volume. This resonance is difficult to apply damping to reduce the peak, or EQ. The nearfield performance targets limit the size of the enclosure and opening.
Next Steps: Optimize for Far-Field Applications
Future goals include:
For far-field-only designs, new challenges arise
The IAR Mouth Source V1 proves that a high-performance, standards-compliant acoustic source can be made using 3D printing and off-the-shelf components. While there are limitations, the design offers a strong platform for future development and upcoming iterations. As a sneak preview- we know we can extend the bandwidth with the "big mouth" design shown below, if we are ok being slightly out-of-spec on the P.51 near-field directivity. And since most interfering talkers are in the >0.5 meter distance, this seems like an acceptable tradeoff.
The prototype mouth achieved improved high-frequency bandwidth with no change in low-frequency performance. A limiting factor was the Helmholtz resonance near 2 kHz, caused by the mouth opening interacting with the front volume. This resonance is difficult to apply damping to reduce the peak, or EQ. The nearfield performance targets limit the size of the enclosure and opening.
Next Steps: Optimize for Far-Field Applications
Future goals include:
- Improving bandwidth and overall response flatness.
- Enlarging the front opening to eliminate the 2 kHz peak and boost high-frequency output
- Increasing enclosure volume to enhance low-frequency response
For far-field-only designs, new challenges arise
- If the mouth cannot be calibrated at MRP, what becomes the reference point?
- What off-axis directivity target should be used?
The IAR Mouth Source V1 proves that a high-performance, standards-compliant acoustic source can be made using 3D printing and off-the-shelf components. While there are limitations, the design offers a strong platform for future development and upcoming iterations. As a sneak preview- we know we can extend the bandwidth with the "big mouth" design shown below, if we are ok being slightly out-of-spec on the P.51 near-field directivity. And since most interfering talkers are in the >0.5 meter distance, this seems like an acceptable tradeoff.
RSS Feed