We are happy to announce availability of a new product ACA – Audio Compression Advisor!

Recording large amount of audio always requires a proper selection of sampling frequency and bit rate to optimize storage and quality of the recording. A precise and more sophisticated approach is to use AQuA Wideband, but when you need a quick advice on what sampling rate and bit rate to use for your podcast, meeting recording, conference speech, audio book or other large amounts of audio signal Audio Compression Adviser is the product that gives you proper parameters for your MP3, OGG, AAC encoder.

Give it a try with this online demo, provide UNCOMPRESSED .WAV (note the online demo accepts files only uncompressed wave files less than 5Mb ), set how much quality you like to preserve in the compressed audio in percentage and click “Advise!” button. After the file is uploaded you will see recommended sampling rate and bit rate for your audio within a couple of seconds!

Contact us to receive ACA for evaluation!

AQuA Music is based on Sevana’s patented technology (AQuA Wideband) for perceptual evaluation of audio signal quality.

When considering digital recording and digital playback, sound quality depends on the range of sound signal, the rate at which it is sampled, and various conversions that occur in sound reproduction system. In lossy codecs like MP3, sound quality is a factor that determines how much of the sound signal data the encoder is allowed to loose in order to reduce file size. In MP3-encoded signals for instance the quality is defined by its bitrate, in kilobits per second (kbps).

The frequency range of sound (in Hertz) which equipment is capable to sample and reproduce affects sound quality. Humans can hear frequencies ranging from about 20 Hz to approximately 20 kHz, so sampling that does not extend far enough will take effect on the sound quality.

Sound signal wave is continuous and has some real value at every instant. The digital quantization of the analogue sound wave means that much of the continuous sound wave is not recorded. The rate at which the sound is sampled refers to the amount of information the detection equipment records for each second of the sound. The higher the sampling frequency, the more accurate the final samples will be. However, there is a mathematical proof that in order to preserve all frequencies in the original sound it’s enough that the sampling frequency is two times higher than the highest frequency in the sound signal spectrum. Thus for example when dealing with speech one can easily use sampling frequency of 8kHz, because human speech spectrum is limited by 4kHz and telephone channel spectrum by 3.4kHz.

When recording a podcast, grabbing music from your CD into MP3, OGG or AAC, uploading an audio book to your iPhone or dumping your favorite tracks on a memory stick to play while driving, we always face the same problem – we want to make the file smaller but sound as good as ithe original. In many cases when one usess MP3 encoder the choice he or she makes is simple – to use 320kbps and that will surely preserve the quality. In many cases it may work, but what if you store thousands of tracks? What if millions? What if you are going to transmit your podcast over the Internet? What if your audio is transmitted over a mobile network? In such cases every megabyte saved turns into real money saving and the lower the bitrate preserving desired audio quality the better performance of your service.

The way it works is very simple. AQuA has a perceptual audio quality model that acts as a very well trained human ear that can hear even tiny changes in audio quality and tell how much quality was lost due to audio processing. AQuA returns two types of scores:

- percentage of quality similarity compared to the original audio
- MOS (Mean Opinion Score), which is a value between 1 and 5

So, the process of choosing the best audio compression parameters (OGG, MP3, AAC compression optimization) when optimizing f.e. bitrate with Lame MP3 encoder is split into the following phases:

1. set minimal bitrate = X
2. set bitrate incremental step = S
3. compress original audio into MP3 with bitrate X
4. decompress MP3 into uncompressed WAV
5. test quality comparing original WAV and uncompressed MP3 using AQuA Wideband
6. if the quality is lower than MOS 5 increase bitrate so that new bitrate = X + S and go to step 3
7. if the quality is higher or equals MOS 5 then we have found optimal bitrate for this audio

Let’s use MP3_Torture_Test.wav taken from this location: http://recording.org/daw-pro-audio/28203-mp3-encoder-torture-test.html

We are going to use mp3opt.bat file to automate bitrate optimization and the result we get is:

“% = 97.59″
“MOS = 5.00″
“Bitrate = 104″
“Passes=13″

So, we preserve 97.59% of original file quality if encode MP3_Torture_Test.wav using Lame encoder with the following parameters:

lame.exe -b 104 MP3_Torture_Test.wav MP3_Torture_Test.mp3

In the post where we took the MP3_Torture_Test.wav from 128kbps gives a very good quality according to the poster, but we could save 24kbps on bandwidth and about 400K on the file size!

lame.exe -b 128 MP3_Torture_Test.wav MP3_Torture_Test-128kbps.wav.mp3

Now it’s time to check how much quality was preserved in the audio compressed with 128kbps, so we use AQuA again:

aqua-wb tst.lic -mode files -src file MP3_Torture_Test.wav -tstf MP3_Torture_Test-128kbps.wav -acr auto -npnt auto -miter 1 -enorm on -grad on -mprio on -tmc on -ratem %m

Here is the output:

Sevana Audio Quality Analyzer – AQuA-Wideband v.5.3.11.720.
Copyright (c) 2009 by Sevana Oy, Finland. All rights reserved.
—————————————————————
test license
—————————————————————
File Quality is
Percent value 97.52
MOS value 5.00

Amazingly, but the quality is the same as at 104kbps! And if it sounds the same (you are welcome to download these files and listen to them yourself) why to spend extra space and bandwidth?

And now another application of AQuA Wideband – audio transcoding. Let’s compress MP3_Torture_Test.wav using Lame with 320kbps keeping in mind that this is a typical choice of those who want to preserve maximum of quality.

AQuA shows us that at 320kbps (MP3_Torture_Test-320kbps.wav.mp3) we preserved

aqua-wb tst.lic -mode files -src file MP3_Torture_Test.wav -tstf MP3_Torture_Test_320kbps_mp3.wav -acr auto -npnt auto -miter 1 -enorm on -grad on -mprio on -tmc on -ratem %m

Sevana Audio Quality Analyzer – AQuA-Wideband v.5.3.11.720.
Copyright (c) 2009 by Sevana Oy, Finland. All rights reserved.
—————————————————————
test license
—————————————————————
File Quality is
Percent value 99.50
MOS value 5.00

almost 100% of audio quality. Let’s now transcode this audio file and check what lowest bitrate we can use and how much quality we’ll preseve from 99.5%

“% = 97.53″
“MOS = 5.00″
“Bitrate = 104″
“Passes=13″

And we got a pretty much the same result from tanscoding. Of course when dealing with uncompressed original audio we have a better chance to avoid compression artifacts, but still it is possible to transcode with a lower bitrate and audibly the same quality. Of course compared to the original uncompressed audio we may get a different result:

Sevana Audio Quality Analyzer – AQuA-Wideband v.5.3.11.720.
Copyright (c) 2009 by Sevana Oy, Finland. All rights reserved.
—————————————————————

test license
—————————————————————
File Quality is
Percent value 97.41
MOS value 5.00

But it is still of a high quality level.

After transcoding we saved 216kbps bandwidth and 70% of space!

What’s in it for me?

Private customers (AQuA Service):

• Optimized audio CD grabbing
• Increased storage on MP3 player, iPod and mobile devices
• More audiobooks to fit the same storage
• More tracks to fit the same memory stick
• Simplified quality based personal podcasts encoding – optimized bitrate and audio size

Corporate customer (AQuA Server):

• Tremendous space saving on large audio files
• Space and bandwidth saving on podcasts, tunes, audio books hosting and streaming
• Optimized audio for transmission over communication channels
• Bandwidth saving on mobile network audio streaming
• Automated audio encoding with pre-defined quality level

Voice quality is one of the main characteristics of speech transmission systems. When analyzing voice quality one must not only consider audio signal degradation caused by transmission over telecom channels, but also specifics of speaker’s voice, conditions of listener’s hearing and variation of these parameters in time.

The most known methods for quality evaluation of voice transmission systems were developed by Telecommunication Standardization Sector of International Telecommunications Union (ITU-T) in the middle of 90-s. Results of this work are presented in Recommendation P.800 (P.830) «Methods for subjective determination of transmission quality» [1, 2]. This document describes conditions for voice quality testing, audio contents, scoring and methods to evaluate results. Typically “Methods for subjective determination of transmission quality” are used to obtain mean subjective quality score according to five-digit scale (Mean Opinion Score – MOS).

Unfortunately P.800 recommendation tests may lead to ambiguous results. Recommendation is warning about comparing MOS scores received under different conditions and consider such approach incorrect. Besides that preforming tests according to P.800 takes a lot of time and requires a lot of testers involved in the process.

In order to move from subjective (MOS) scores to objective ones and to automate the quality measurement, ITU-T has developed the P.861 recommendation, which is based on low level quantitative measurements [3]. Recommendation P.861 is a follow-up of PSQM method (Perceptual Speech Quality Measurement), developed by KPN Research and devoted to objective analysis of speech codecs performance with a low level of degradation.

However, it is impossible to utilize PSQM for evaluation of work of a real communication system because the method does not consider all the important factors influencing human perception. Among these factors are delay, jitter, packet loss as well as signal level clipping.

In February 2001 ITU-T has issued another recommendation ITU-T P.862 [4], which describes a more advanced algorithm for voice quality testing – PESQ (Perceptual Evaluation of Speech Quality). The algorithm includes level and time aligning, human perception and cognitive modeling. Due to these additional operations the approach considers signal amplification/ attenuation in a communication system, time delays and jitter as well as spectrum bands, which are the most significant for human perception. Based on cognitive modeling PESQ also recalculates objective quality score into MOS values.

A disadvantage of PESQ as well as other similar algorithm is the fact that they are based on comparing of two signals: original and transmitted through a communication system. This approach may create a range of difficulties connected with setting and preforming voice quality testing. One requires to arrange signal recording on both sides of the telecommunication system as well as records transmission to the test system. Besides this real time quality monitoring in such approach appears quite difficult as well.

In order to solve the challenging issues mentioned above ITU-T has developed a new recommendation P.563 [5] introduced in May 2004. This recommendation determines algorithm for evaluating speech quality by listening to communication sessions. The algorithm takes into account single-side distortions, speech trunk parameters, noise and speech naturalness. Developers of P.563 call attention that P.563 does not provide overall quality estimation of speech transmission. Distortions driven by delays, echo, loss of loudness and everything related to two-sided interaction cannot be taken into consideration by this method.

It’s widely thought that P.563 provides a high level of correlation between automated and expert quality scores. However, simple tests based on ITU-T sound database for codec testing [6] may raise some doubts about the consistence of the algorithm provided together with its description.

Table.1. Comparison between results of P.563 and expert estimations

The problem discovered in the distributed P.563 algorithm implementation required development of an alternative solution. Further down one can find one of possible solutions that is implemented in Sevana NIQA (Non-Intrusive Quality Analyzer).

NIQA’s (Non-Intrusive Quality Analyzer) approach is based on a database of trained etalons called associations. Each association corresponds to a group of files that have close expert estimations of sound quality and common set of reasons for sound quality degradation. For each association NIQA calculates and stores a distribution of parameters’ values.

Basic algorithm showing how NIQA obtains sound quality scores is represented on the picture below.



NIQA - non-intrusive voice quality testing software

When loading sound signal the system excludes all fragments with low energy level (according to threshold). The excluded fragments correspond to “absolute silence” and are considered irrelevant for obtaining sound quality score.

At the next phase the signal is split into frames used in voice activity detection algorithm (VAD). The system calculates energy values for each frame what increases accuracy of VAD. With the help of VAD algorithm the signal divides to active and inactive components that are processed separately. The system builds level histograms for both active and inactive signal components.

By discrete cosine transform (DCT) the system obtains signal spectrum and checks the active components frames for DTMF presence and then excludes the frames that are similar to DTMF from further processing.

Next stage applies the first level of psycho-acoustic model to the signal spectrum. This model checks different types of masking (including pre-masking and post-masking). According to clear peaks of spectrum energy the system splits the signal into tone and noise components.

Second level of psycho-acoustic model performs energy normalization of the signal – energy levels are transformed into loudness levels at 1kHz. Third level of psycho-acoustic model transforms loudness levels into several detectable grades of loudness that allow to ignore sound signal changes, which are not recognized by human ear.

The next step is to split signal spectrum into bands that are critical to human ear perception and calculate parameters both on and out of the bands. Based on the computed signal parameters the system selects most similar associations from the database and performs matching. According to selected associations the system determines how much each of them influence the overall quality and then generates the final voice quality score as a combination of scores for selected associations and according to correspondent weights.

Sevana NIQA Testing and Evaluation

Sevana NIQA has been tested utilizing the same ITU-T speech database that is used for conformance testing of P.563 algorithm. In the tests we used a total of 376 English language recordings. All recordings were sorted into 4 groups depending on their MOS scores (represented in the documentation attached to the sound database). For all groups of recordings we determined average expert scores and average NIQA scores (Table 2). In order to illustrate comparison with P.563 we also calculated average errors for P.563 and NIQA scores for the same tests.

Table.2. Comparison of NIQA scores against expert estimations

The results clearly show that NIQA allows receiving much higher accuracy between generated quality scores and expert estimations than P.563. NIQA scores are less precise only for records with very low MOS scores (in the range from 1 to 2). In all other cases NIQA provides 2-3 times higher quality scores precision compared to MOS values.

References

1. Methods for subjective determination of transmission quality // ITU-T Recommendation P.800 / http://www.itu.int/rec/T-REC-P.800/en

2. Subjective performance assessment of telephone-band and wideband digital codecs // ITU-T Recommendation P.830 / http://www.itu.int/rec/T-REC-P.830/en

3. Objective quality measurement of telephone-band (300-3400 Hz) speech codecs // ITU-T Recommendation P.861 / http://www.itu.int/rec/T-REC-P.861/en

4. Perceptual evaluation of speech quality (PESQ): An objective method for end-to-end speech quality assessment of narrow-band telephone networks and speech codecs // ITU-T Recommendation P.862 / http://www.itu.int/rec/T-REC-P.862/en

5. Single-ended method for objective speech quality assessment in narrow-band telephony applications // ITU-T Recommendation P.563 / http://www.itu.int/rec/T-REC-P.563-200405-I/en

6. ITU-T coded-speech database // Supplement 23 to ITU-T P-series Recommendations / http://www.itu.int/rec/T-REC-P.Sup23-199802-I/en

Copied from: http://www.sevana.fi/non-intrusive-voice-quality-testing-software.php

Non-intrusive voice qualiy testing does not require a reference signal and can work with real communications data in real time.

Non-intrusive techniques provide a possibility to test and monitor a greater amount of communications and therefore obtain a quite reliable information about the networks (VoIP, PSTN, GSM, CDMA, LTE) quality.

Today Sevana announces availability of new software for voice quality testing for customers’ and partners’ evaluation. Be among the first to test the new approach to voice quality testing and monitoring!

This software is delivered only upon request.

Great news about use of AQuA software – now you can easily test voice quality in your or any other mobile network just by using your Nokia N82 mobile phone and Sevana AQuA software. Here is how you can do it.

First of all do you know that  you can record phone calls you make with your Nokia N82 phone? During a phone call just hit Menu on your keypad, select Applications > Media > Recorder. Select Options > Record Sound Clip to record your current phone call. Now you have an audio file for the so called non-intrusive voice quality testing. However, AQuA is the means to do intrusive voice quality testing and therefore we need a reference audio file. There are a couple of simple ways to test voice quality in your mobile network:

1. If your operator allows uploading a greeting audio file into your voice mailbox then you just need another phone or SIM card:

Step 1: Upload a reference audio (you can also use the speech model audio file provided with AQuA for most accurate voice quality testing or just any audio) to your mobile operator facilities as your voice mailbox greeting message (that will be the reference audio) and set your connection to forward all calls to your voice mail when  your phone is unreachable or turned off. Now switch your phone off – you have just created an answering machine to test voice quality of your mobile network!

Step 2: Use another SIM card of the same or other mobile network (depending whether you like to test voice quality in your mobile network or the quality of a voice call from one mobile network to another), insert it in your phone and dial the phone number of the SIM card you used to store audio to your voice mailbox.

Step 3: Your call will be forwarded to your voice mailbox and you will hear the greeting message you uploaded. Record the call as described above, transfer the audio file to your computer and use AQuA to test the original file quality against the recorded one and you will receive objective estimation in percentage of quality, Mean Opinion Score (MOS) and PESQ-like values.

2. You can use an ordinary answering machine (the best digital, so you can upload digital recording into it) in the same manner and plug it into a phone socket of a fixed line. This way you will test voice quality between your mobile operator and fixed line (PSTN) provider.

3. You can use a mobile modem with voice feature enabled (to create a mobile answering machine) and make phone calls from your N82 to it thus testing voice quality between two mobile devices.

Do you have a virtual number and like to test voice quaity between your moblie network and virtual number provider? just set it to forward calls to voice mail and make sure that the greeting message is the same as your reference audio.

Special issue of this post:

- you don’t have to buy AQuA for doing these tests, just contact us and we’ll do testing ourselves providing you with CSV or XML file with the test results

Good luck with your tests!