Audio Quality in Online Learning
From WikEd
[edit] Descriptions, definitions, synonyms, organizer terms, types of
Audio, one of the core delivery methods of both synchronous and asynchronous online instruction, is rarely critically evaluated in terms of its quality. Students often express comments regarding the audio, such as, "too loud," "too soft I can't hear you," "it's distorting," "the echo is bad," and others. Is this identifiably poor quality audio an inconvenience or does it have an effect on the learner? Audio problems are often attributed to the growing pains of distance learning technologies, the fault of the network, and slow computers. Are these the only places to look for improvement?
The broadcast and recording industries have used specifications for audio that provide labels such as "broadcast" or "CD" quality that point to quantifiable measures in two main parameters of the audio recording process. These two measurements are sample rate and bit rate. CD's have a 16 bit, 44.1 khz sample rate signal. Today's internet connections require the use of compressed audio to reduce download times and support any sort of live communication. However codecs, the COmpression and DECompression software process for delivering audio, can deliver near CD quality audio at 128 kilobits per second (kbps). This data rate is easily supported by most broadband connections in the home and certainly over any ethernet network topology that has been used in the last ten years. Even in live online classroom environments such as Elluminate or Macromedia Breeze where lower audio bit rates are used, the audio is capable of quality equivalent to AM radio and at worst telephone quality.
This leads us to the question of why is internet based audio still lacking in quality and why are there complaints? The work of computer scientists at University College London began looking at the audio utilized in online learning environments to answer these questions of the quality both in its effect and cause.
Hardman, Sasse, and Kouvelas found in 1998 that, "Packet loss, multitasking in the workstation, and acoustic problems are the three major causes of impaired audio quality." In other words, slower, unreliable network connections, underpowered computers on the students' and teachers' desks, and physical concerns in the recording space and equipment used were the core reasons for bad audio. It was apparent that bad audio existed, the students were reporting it. Hardman, Sasse, and Kouvelas's work sought to determine how the lower quality audio was effecting the students during online class sessions. They chose to compare subjective reporting of audio quality by the students through a 100 point scale rating against physiological responses measured by blood volume pulse (BVP) and heart rate (HR), a way to measure stress or the 'fight or flight' reflex.
Hardman, Sasse, and Kouvelas found that subjective perception of degradation did not directly correlate with stress responses measured with blood volume pulse and heart rate. In general as conditions degraded due to any conditions (network, computer, mic, etc.) both the user reports suggested there was problems and the BVP and HR data suggested increased stress. However the direct linking of the factors causing degradation did not match. The users identified that the worst audio conditions happened when there was 20% packet loss causing choppy and interrupted audio. While the stress measurements placed the bad mic or excessive volume conditions as the biggest cause of stress.
The computer workstations and network problems have been tackled from a variety of sources. The compression of audio has experienced improved quality while decreasing bandwidth and file size requirements. The work of computer scientists have created a more robust and stable internet with better error correction in file transferring even under less than opportune conditions. The conditions of 20% packet loss in internet transmission are quite rare. It is safe to say that the network itself is not a prime contributor to poor audio quality. Today's desktop and laptop computer, if properly maintained, have no problems with audio. This leaves excessively loud, echo and bad mic conditions as the cause of user dissatisfaction with audio quality. As Hardman, Sasse, and Kouvelas state it concisely, "In the subsequent nine years, different approaches to internet error correction, significantly faster and powerful computers available the masses have greatly improved issues with packet loss and multitasking. This leaves the less quantifiable concern of acoustic problems."
The problem is not a computer/network issue anymore, it's a primarily a human issue of training and technique. In addition, a good mic helps. What is "good" will be discussed next.
[edit] Application in online classrooms
Given the identified increase in stress levels demonstrated through BVP/HR data and remaining reasons from user reports, it can be safely stated that poor quality audio can have a hindering factor on the learning environment through raising stress levels. While it is difficult to dictate to online students what microphones to buy, instructors and institutions can do their part to improve the audio they deliver. While it is beyond the scope of this article to discuss the entire audio signal chain, acoustic treatment of rooms and their potential optimizations a few key points can be made. The book, "Sound Studio Construction on a Budget" by F. Alton Everest listed in the references is one of the best sources available for room optimization for audio recording and reproduction.
As described above, the problematic situations traditionally associated with poor quality audio have been found to be not the fault of the internet or the computing technology. Increased bandwidth and evolved software/hardware have solved most of the problems experienced ten years ago. As Watson and Sasse identify, the problem is in the user's set up and control of the audio environment.
Volume discrepancies, poor quality microphones and echo have a greater impact on the user, meaning that it is possible to have perfect transmission from a network viewpoint, but still have poor quality audio from a user’s viewpoint. The solutions envisioned mainly involve raising and improving user awareness, both of what the problem is, and how to solve it. These can be low-cost solutions - a huge amount of people-support should not be required once audio tools are better set up to support non-expert users.
[edit] Understanding Improvement of Online Compressed Audio
To squeeze audio down to a size that permits transmission to multiple class members, the computer needs to throw away sonic information. The problem is that background noise, handling noise, breaths, and mechanical noise all are "heard" by the microphone. The computer gives equal processing to the unwanted sounds and the teacher or student's voice. The data spent on reproducing the unwanted sounds takes away from the potential quality of clean voice reproduction if better microphones and related equipment, and a acoustically treated room are used.
[edit] Environment
The environment is extremely important as was witnessed in the design of all the early distance learning facilities in the 1990's. Specialized rooms were often constructed to create a "studio" space with special seating so everyone could be seen by the video camera, TV monitors, mics and sound treatment on the walls and often ceilings. F. Alton Everest states that the goal of this construction and treatment was speech intelligibility. Comparatively, a one person broadcast room is fairly easy to deal with today. The person speaking should be in a quiet room that doesn't have many resonances or short echo/reverberations. If the room has hard surfaces on all walls and floors there will be problems. Our ears filter this "ringy" sound, but mics and computers are not that smart and attempt to reproduce everything. Office dividers are excellent absorbent panels that can correct this. Anything soft and squishy can help. Staying clear of corners also helps. This simple test is clap your hands once and listen, you will hear a short decay of the sound, but any chattering, short echoes and harsh rings need attention.
[edit] Mic choice
[edit] Giving Up Video When Possible
Despite the romance of seeing yourself and others on the computer screen, the jury is out whether seeing each other's faces has a significant impact on learning. By observing the following CTER chat transcript it can even be said that video of only presenter faces may be a distraction. This is taken from the middle of a lecture in Elluminate. You will see the professor starts the ball rolling on distracting comments about the TA (referred to as "TA lecturing" rather their name) looking like an astronaut. You'll see "student 13" tries to stop his fellow classmates from the name calling and teasing. There was no compelling reason to use video for this lecture.
student 1 Cter: true, that is my fear
the professor Cter: If they don't you will have trouble getting them to pay attention to you
student 1 Cter: oh, so our results don't have to be finalized?
the professor Cter: Doesn't TA lecturing look like he is an astronaut speaking to us from a space capsule
student 1 Cter: oh, nevermind
student 1 Cter: (yes! totally!)
student 2 Cter: ha
student 3 Cter: i was thinkin StarWars
student 4 Cter: yes...you have it
student 5 Cter: haha yes
the professor Cter: Yes, Sarah, you should have final results
student 6 Cter: Perhaps shifting focus from "why requested" to "why you believe this is important" will help?
student 1 Cter: ok (crossing fingers)
student 7 Cter: We just need to hear those annoying mic beeps.
the professor Cter: Let's take a vote--how many people think TA lecturing looks like an astronaut?
student 2 Cter: (raises hand)
student 8 Cter: yes
student 6 Cter: Darth Vader
student 1 Cter: yep
student 8 Cter: its really funny
student 5 Cter: astronaut
student 9 Cter: yep
student 10 Cter: yes
student 11 Cter: Is he lipsincing?
student 3 Cter: haha!
student 12 Cter: poor TA lecturing
student 4 Cter: I think he's drinking tang
student 13 Cter 1: let him speak guys
student 2 Cter: how's he going to come back? they delayed the shuttle launch!?
the professor Cter: sorry, TA lecturing
student 5 Cter: Can we get a full view instead of a head shot
student 7 Cter: How can he read and think his own thoughts at the same time?
student 2 Cter: astro-alien powers
Video has its purpose from demonstration and observational situations. If it can be omitted, the bandwidth saved can go toward significantly better audio.
[edit] Evidence of effectiveness
Hardman, Sasse, Kouvelas and Watson have provided the most solid data supporting the problems with and directions toward solutions. Their work has told us to foster better user control and knowledge of their audio enabled workstations, microphone technique and recording environment. When audio is done well it hopefully is transparent. This user awareness of problems can apply to video conferencing and production as well.
Work done on the campus of the University of Illinois provide many demonstrations of quality. The Apreso classron audio/video/slide capture system in the Seibel Center, the Distance Learning Media Center in the College of ACES, and the audio production originating from the College of Letters and Sciences are recognized leaders on the campus. Their evolving work and documentation of best practices is a work in progress at the following address: UIUC Multimedia Users Group
[edit] Critics and their rationale
There doesn't seem to be any arguments for implementing poor quality online audio. While budgetary issues may cut or not offer funding for the needs identified above, there doesn't seem to be any good reason to not improve audio.
[edit] Alternative explanations due to Diversity considerations
Diversity issues raise two points with online audio. The first is a concern over access to mutltimedia computers. This concern may be across socio-economic, geographical constraints or both. While a full treatment of the topic would be a book unto itself promising work is being done in the area. The One Laptop Per Child project has begun to release a $200 multimedia capable, internet ready computer to distribute through governments in underserved countries. The machine is rechargeable with a foot or hand generator, is readable in bright sunlight, utilizes a new network wireless infrastructure that works in remote areas, can take pictures, videos and record and play sound. The machine is quite amazing. While not the studio quality audio possible, it works well.
The second diversity consideration with online audio is one of accessibility for people with special needs; specifically the hearing impaired. The Americans With Disabilities Act prescribes various practices for providing alternate design and publication methods for online content that is audio based. While adapting all content is a technically feasible proposition, the funding is limited and not recurring currently to pursue production systems that would be capable of providing all audio content in alternate methods at UIUC. According to Paul Riismandel, the leading multimedia producer for the College of Letters and Science at UIUC states that they currently rely on the student that require accessibility accommodations to contact the instructor who then works with the multimedia producers to have content transcribed and converted to other deliverable sources.
[edit] Signed “life experiences”, testimonies and stories
At the Siebel Center for Computer Science at the University of Illinois, Urbana-Champaign, there are five classrooms equipped for audio and video capture. For the most part, video has proven to be relatively easy for us; either there is a video signal, or there isn't - we've had few problems with the quality of the signal. Audio, on the other hand, has proven to be a challenge for many reasons. The microphones available to instructors are professional quality wireless lapel microphones that feed into the classroom's sound system, which is heard in the room via ceiling mounted speakers. That same audio feed is also routed to an encoding computer in the classroom, as well as sent over the fiber network to a central control room, where it is then routed to a back-up encoder. We've received few complaints from students in the classrooms, but we have had complaints from the remote students viewing the recorded lectures online. One of the main reasons why it's so difficult to troubleshoot problems with audio is that audio can be subjective - what sounds acceptable to one listener may not be to another. Also, there are usually multiple points in the signal chain where the audio signal can degrade or become distorted. Not to mention that the signal at the source can be poor if the instructor is not wearing the microphone properly. Then there's the issue of remote students wanting to be able to hear any classroom discussion or questions from the on-campus students. It's unreasonable to expect all the on-campus students to wear microphones, of course, and we've tried implementing wireless handheld microphones that the students can pass around, but that has been met with limited success. Some students are not comfortable speaking into a microphone, and having to pass one around can disrupt the natural flow of the discussion. One solution, while less than ideal, has been to mount a "shotgun" microphone in each of the classrooms high enough to minimize the capturing of background noise. These microphones are very directional and do filter out some of the unwanted noise, but there is no practical way to filter out all of the unwanted ambient sound. So far, this solution has proven to be largely acceptable by the remote students viewing the recorded lectures online. -Scott Cimarusti
Audio quality for us has been largely a function of budget. Though we have a production department that doesn't a lot of AV work at FIDM, the eLearning team has had to get it's own hardware and gear to do much of the podcast and streaming audio recording on their own so we wouldn't have to wait in line for a "sound guy." We at first purchased a unidirectional microphone and a wireless set in order to get the basics done. When you are including the audio in a multimedia piece, you have control over the file and the quality/filesize but we kept running into problems doing guest speaker recordings since they would have problems speaking directly into the microphone. (This was when we were out of 9 volt batteries for the wireless lavaliere mic or it was used by a second speaker). Moral of the story, get good equipment. An additional lav and an omni-directional is in our future. - Dave Melone
Dave, thanks for the comment. Take a look at a multipattern mic offering omnidirectional, figure 8 and unidirectional characteristics. They all have there place. The Snowball mic shown above is a great multipattern USB mic.
[edit] References and other links of interest
- TOWARDS AN EFFECTIVE USE OF AUDIO CONFERENCING IN DISTANCE LANGUAGE COURSES, Language Learning & Technology
http://llt.msu.edu/vol8num1/hampel/ January 2004, Volume 8, Number 1, pp. 66-82
- Proceedings of the international ACM SIGGROUP conference on Supporting group work, Phoenix, Arizona, United States, Pages: 160 - 168, Year of Publication: 1999, ISBN:1-58113-065-1
- Everest, F.A. “Sound Studio Construction on a Budget”. McGraw Hill Inc. 1997. New York, USA, pp. 118-120.
- Fraunhofer Institute http://www.iis.fraunhofer.de/fhg/iis/bf/amm/
- Hardman, V., Sasse, M. A., & Kouvelas, I. (1998). Communications of the ACM; successful multiparty audio communication over the internet: The internet was once perceived as a computer network used by researchers to transfer files and send text messages. today, more users are becoming aware of its potential as a general communication networkAssociation for Computing Machinery, Inc.
- Watson, A., & Sasse, M. A. (2000). The good, the bad, and the muffled: The impact of different degradations on internet speech. Proceedings of ACM Multimedia 2000, , 269-276.
- One Laptop Per Child Project, http://www.laptop.org
- ADA information from meeting of the UIUC Multimedia Users Group Steering Committee, April 2, 2007.
