The E-Learning Curve Blog has moved!

You will be automatically redirected to the new address in 10 seconds. If that does not occur for some reason, visit
http://michaelhanley.ie/elearningcurve/
and update your bookmarks.

Thursday, August 13, 2009

Podcasting for E-Learning: Digital Audio Basics

Really exceptional podcasts transmit a sense of "being there."

Well-designed, planned, narrated and produced audio content has the facility to engage the audience and transport the listener, providing the right cognitive environment for learning or entertainment or whatever goal the podcast seeks to achieve. As the old story goes, “there are better pictures on the radio.” Denis Nowlan, Deputy to the Controller for BBC Radio 4 describes this as

the wonder of radio [and sound] …the conjuring of magical sounds, Prospero-like, out of the silently teeming air. And the sense of joining, instantly, with a vast community of listeners.

(2005)

Of course, this cognitively experienced sense of “being there” is a completely artificial construct, none more so when you consider audio podcasting. The journey the podcaster takes us on is based on nothing in nature; that is, a narrative transmitted as binary data: electronically-mediated zeros and ones.

I will discuss this phenomenon in detail next time, but for now, read on…

Digital audio is actually very similar in concept to motion film, where a rapid series of still photographs make up a “motion picture.” Digital recording reproduces audio signals by taking many still “pictures” – or samples - of an audio waveform and then reconstructing the waveform for playback digitally (see Figure 1).

audio sampling

Figure 1. Digital Sampling of Audio Waveform
[Click to Enlarge]

In film, the number of still images (frames) we move through the projector per second improves the smoothness and quality of picture we see. Similarly, the size of the film's light-sensitive crystals in each frame determines the visual resolution of the image - and we all know movies look better in 70mm as opposed to the 35mm or 16mm film.

In digital audio, an analog signal is typically sampled 44,100 times a second. Each sample is created by taking a "snapshot" of the amplitude (or strength) of an analog audio signal at a particular moment in time. and converting that amplitude into a binary number.

The number of samples contained in each second of audio is called the sample rate. The greater the number of samples per second, the greater the resolution of the audio signal.

Let’s return to the film analogy: the faster the film rate through the camera, the better the ability of the film to pick up motion. For example, if you crank a film camera too slowly, then motion past a camera is not smooth since the frame rate does not capture enough frames in a given amount of time to stop the motion from appearing to jump or jerk - too much action happens in front of the camera "between frames" for the motion to appear natural.

Time for an example: The resolution of an audio CD is 16 bits and cd the sample rate is 44.1KHz. This means that the audio is sampled 44,100 times a second and a series of 16 zeros and ones is used to define the amplitude of the waveform at each of those 44,100 points.

This resolution – commonly called “CD Quality” and should be your baseline for recording a podcast. When recording voice-over narration, capture all of your audio at 16 bit, 44.1KHz, in mono using the WAV format in Windows, Au format for Unix-based systems (like Linux), and AIFF for Apple Mac.

More...

__________

References:

Nowlan, D. (2005). Radio: where the pictures are better: Denis Nowlan speaks for the medium which leads you by the ear. [Internet] Available from: http://findarticles.com/p/articles/mi_m0KZH/is_1_19/ai_n15928065/ Accessed 12 August 2009.

--

2 comments:

Unknown said...

Wrong: In digital audio, an analog signal is typically sampled 16 or 24 times a second – that is, 16 or 24 bits of data.

They are sampled 2^16 times a second or 2^24 times a second respectively.

Michael Hanley said...

N,
Actually we were both incorrect, audio is sampled 44,100 times per second, at a resolution of 16 or 24 bits per sample.

Thanks for the FYI tho' - it's enable me to correct the error.

Regards,
Michael
--