PESQ. POLQA. AQuA. Feature Comparison.
Hi,
I thought of sharing some of our discoveries comparing standard technologies for end-to-end voice quality assessment (PESQ, POLQA) and our in-house developed AQuA technology. Here is what we have figured out:
In fact AQuA can synchronize reference and test audio in time and signal level domains with variable delays shorter and longer than one second.
AQuA perceptual model naturally can work in all bands thus providing audio quality analysis from 8kHz up to 192kHz sampling rate.
AQuA unlikely PESQ or POLQA has ability to configure perceptual model by parameters profiling. To our experience this helps achieving more accurate MOS scores depending on actual environment/network.
AQuA copes with P.800 standards and this means that if reference audio and test audio are identical their comparison MOS score will be equal 5.
With the help of our clients we also found several features that significantly differ AQuA from PESQ and POLQA:
- AQuA perceptual model is language independent
- AQuA provides MOS score for audio with duration starting from 2s
- AQuA provides MOS score for audio with duration longer than 30s
- AQuA can analyze stereo audio and higher sampling rates (like it was mentioned above)
- Parameters profile allows configuring AQuA perceptual model to give more precise MOS for different environments (codecs, PSTN, GSM, VoLTE, VoIP etc)
All technologies may have advantages and disadvantages, but if you are planning to enable end-to-end (or single-ended) voice quality testing we welcome you to contact me directly or through our web site. Please read more about AQuA at https://sevana.biz/products/aqua/ and https://sevana.biz/products/aqua/technology/
If you have questions related to voice quality analysis you are welcome to join our Voice Quality Assessment Tools LinkedIn group? and Google Group.
Would appreciate your questions, comments!
Regards,
Valeri