Country Selection and IT Outsourcing - Talk

An open invitation to "Country Selection and IT Outsourcing" - all welcome. Please forward.
Speaker: Allen Higgins
When: Midday (12:00-12:30), Tuesday 24th January, 2017.
Where: Lecture Theatre 1, UCD Smurfit Graduate School of Business, Blackrock, Dublin, Ireland.

In this talk we ask; are offshore IT outsourcing (ITO) and IT mediated business process outsourcing (BPO) linked with social conditions in supplier countries? We compare international outsourcing indices produced by consulting firms, with country-level social data obtained from international data sets. We find that two country attractiveness factors (financial attractiveness and people & skills) are correlated with three international data sets (child mortality rate, life expectancy and unemployment). We conclude that higher commercial attractiveness raking is associated with lower socio-economic performance; in effect, poverty makes some destinations appear more financially attractive.

In light of these findings further investigations are proposed to better understand the impacts of offshore outsourcing activity in destination regions. This involves investigating linkages between country attractiveness for ITO/BPO activity and country social performance and how they change over time; and to seek additional sources of reliable international economic data for ITO/BPO activity.

Previously presented at the 27th Australasian Conference on Information Systems 2016.

Download a copy of the paper from ResearchGate. https://www.researchgate.net/project/Impact-of-IT-Outsourcing

Editing audio for a podcast

Audacity 3 (https://www.audacityteam.org)

Audacity is a DAW, a Digital Audio Workstation. Audacity is also an open source project. It is a highly regarded and extremely capable software environment for sound engineering. Edit your audio recording (in post) to tidy things up. If you have a good quality original recording then the Edit makes it better. If your original recording has issues, then the DAW may help you make corrections like: balance the volume of different voices, standardise the levels, reducing pauses, 'emms and ahhs', delete different takes, filler speech, background noises and any generally unnecessary audio. Audacity can also be used to record directly to the project if needed for voice over and bookend statements.

Audacity Effects may improve sound quality, audio clarity, apply new elements and create the finished file. You may applying effects like normalise or amplify, compressor, noise reduction, high pass, low pass, equalise, add stings, fades, and eventually, and then ultimately, export the finished project as a releasable mp3 file.

A screen shot of an Audacity project

The screenshot above illustrates what waveform levels might look like for your recording. Above all else  trust your ears to judge the quality of a recording when you listen back to it.

Editing Audio for Podcasting with Audacity

Save Save Save - the Project file and folder

Everything lives in .aup3.

You are strongly advised NOT to work on your active project on an external USB stick/disk, networked storage or cloud storage as it is unlikely to be fast enough for satisfactory recording and editing. To avoid data loss save the project frequently. Make a local disk copy for project editing as Audacity generates very large sets of project edit/state files (online cloud drive services may not unpack or stage quickly enough to support Audacity's project edit/save cycle). To share an Audacity project with others you will need to share the project aup3 file. 

n.b. older versions of Audacity stored data in a project _data folder.  If you move to Audacity 3 there will no longer be a _data folder for each project (yay).

And save save save save frequently. 

And occasionally 'save' then 'close' the project window to force the project to compact safely. 

Reopen and continue working.

Importing audio

Create a new project and save to disk.

    File > Save Project > Save Project (cmd S)

Import previously recorded audio (an mp3 or wav file). 

    File > Import > Audio

Save

    cmd S 

An Audacity project with at least one audio track will resemble the screen shot above.

Play, pause and even record new audio to the current project. 

The playhead is the vertical line displayed over the waveform. All tracks will play through the audio device unless muted or one is soloed. The keyboard space bar toggles the play/pause function from the current playhead.

Once you start importing audio (particularly high quality poly WAV recordings), the Audacity project file (as of version 3 this will be the *.aup3 file) can become very large very quickly (anywhere from 3 GB to 36 GB). 

Multi-track, stereo, mono

A typical audio project file contains synchronised separate tracks representing each microphone or audio source. Usually edit the project configured with Tracks>Synch-lock "on" so that the separate audio tracks remain synched. A basic workflow for post-processing each track separately:

• Loudness Normalisation (-20dB LUFS)
• High Pass filter (6dB at 100Hz) 
• De-Clicker
• De-Esser
• Noise cancellation (if needed)

Some useful editing techniques

Quick select, cut and paste

The Audacity toolbars, click and drag to rearrange or float

Choose the Selection Tool (F1) to edit your audio track directly. Click-drag to highlight parts of the audio waveform. Use the cursor to select the waveform to copy (cmd C), paste (cmd V), duplicate (cmd D) cut (cmd X) or delete (del, backspace or cmd K) along the timeline. The edit commands are also accessed via the menu.

    Edit > etc.

Automatically suppress background and soft sounds like breathing

Use Label Sounds to automatically search for and label silence (Audacity manual)

Select Silence and suppress e.g. Silence Audio or manually e.g. Select and Normalize -35dB

Multitrack 

    Tracks > Sync-Lock Tracks (on/off)

You can organise the structure of a recording by using new tracks for sections of audio (to organise the large scale edit).

As of Audacity 3 you can now grab the track label bar when the hand icon/tool appears. This is a really handy :-) approach for aligning track segments or to move a split selection from one track to another.

Toggle the Sync-Lock Tracks setting either synchronise (ON) edits across the tracks in parallel, or (OFF) move tracks relative to each other. This is useful if for example you want to overlap, reorder, or otherwise change the structure of the recordings, e.g. adjust fades, add transitions, change the order of speech.

Large scale edit

Your listeners want a well edited informative podcast so remember, less is more. You will have to listen (and re-listen) to your material and decide what to keep or delete. Think about the length and structure of the recording. You can change the order. You will delete stuff so do it. Concentrate on large chunks, the structure of the talk. Don’t worry about breathing, coughs and minor noises just yet. 

Small scale edit

After the large scale edit more or less done, spend time on the small scale edit. This involves zooming in and fine-grained cut and pasting. But think before acting, is it really necessary to remove every single intake of breath? 

For example: cut long gaps, background noises and any generally unnecessary audio. You might also cut some of the egregious 'emms and ahhs' and nervous filler speech we all use. A caution though, the small scale edit can become a journey without an end, or at least a very long journey.

Note: "Sometimes you will not want to close the gap and yet still remove the noise, for instance if somebody breathes loudly. For such occasions copy a section of audio during which nobody is speaking (e.g. at the start of the recording) and paste it over the unwanted noise. This will preserve the natural pacing of speech." (source: Cook & Holdis, CEU Podcast Library - link)

Fade-in Fade-out

Apply fade-in and fade-out when you want the speech to start or end smoothly.

A useful Audacity EQ setting

Post-processing - Loudness Normalization 

Start out by applying loudness normalization (default -23 LUFS) separately to each track - why separately? Each track has its own average. I think normalising across all tracks at the same time biases the average gain to the loudest track's level. The same thing happens with Normalize but you can use that to your advantage to selectively boost or reduce selected loud or quiet segments.

Post-processing - noise reduction (if needed)

Select a several second sample of just the noise you want to remove.

Post-processing - low pass / high pass and other ways to reduce echo?

What to do if the original recording has transient audio peaks (annoying clicks, chirps, squeaks etc)?

The following has worked for me.

Apply the low pass filter at as low a cut-off level as you can get away with e.g. 1,100 Hz, 12 dB roll-off

Then apply click removal (200/40)

Another alternative is to apply the RIAA filter using the Filter Curve EQ effect. This shifts the vocal emphasis towards the bass without over-muffling the sound. Reducing the intensity of the higher frequencies allows the listener to focus on the mid-low vocal range and deemphasises the intensity of any annoying high-end audio reflection that gets captured so easily when using a condensor mic in a room without sound treatment.

Then EQ->Normalize->Compressor->Normalize (can repeat if needed) as follows...

1. Post-processing - Graphic EQ (equalization)

Base boost plust treble boost

2. Post-processing - Normalize

Apply default (-3dB) - the waveform will push over 0.5 but keep away from 1.0

3. Post-processing - Compressor

Select -19 dB or choose a threshold between default -12dB down to as low as -30dB 

Select noise floor, around -50dB

Select ratio 4.7:1 or vary, typical ratio, between 2:1 - 3:1 - 5:1

Drop attack time (0.2s) and release time (2.0s) to maximise compression applied

These settings end up boosting the volume (visually the waveform moves closer to 1.0) making it too loud and noisy so you'll need to normalize (or amplify) again...

4. Post-processing - Normalize (again)

Apply default (-3dB) or you can apply Amplify -3dB, for the same result. 

The desired result being for the waveform peaks to re-balance back to around the 0.5 level.

Improving a Zoom track?

Loudness Normalization > AUMatrixReverb > ClipFix > Normalize 3 > Amplify 3

Apply Noise Reduction if a suitable noise sample is available

Apply De-Esser and De-Clicker if needed

Overall, peak levels to be no more than 3dB

AUMatrixReverb settings may help to obscure small room reverb

Alternatively or in addition you might get some improvement by applying the AUBandpass Filter 

AUBandpass Filter

Finish up

Export the mp3

Finally, create the finished product, by exporting to an mp3 file. 

> File > Export > Export as MP3

This mixes all tracks in the Audacity project down to a single track. 

Use settings similar to the following to export your audio file.

Quality checks:

• Is the intro/outro music at a suitable level?
• Does the intro/outro transition smoothly to/from the spoken audio?
• Is the spoken audio at a suitable level? (i.e. no clipping, waveform amplitude averages around 0.5 on linear scale)
• Is the speed of speech natural?
• Large scale edit - remove fumbled speech, removed confidential disclosures, cut out a duplicate or passages that were off topic, etc.
• Small scale edit - remove ticks, emms, ahhs, sighs, distracting background, etc.
• You wouldn't normally have backing music running throughout a podcast, but in some cases, for example, background room noises leaking into the audio, so you may decide to use low volume background as a filler/distractor.

Useful links

An Audacity tutorial on mixing a narration with background music
https://manual.audacityteam.org/man/tutorial_mixing_a_narration_with_background_music.html

Refer to the Audacity manual online (e.g. the track panel description/explanation is at manual.audacityteam.org)

An Audacity user's simple tutorial and workflow activities:

https://podcasts.ceu.edu/how-edit-your-podcast-audacity-step-step-guide

Using the Compressor in Audacity

https://youtu.be/oztvmA7adDc

Using Noise Removal in Audacity

https://youtu.be/vCx4m5dOA9A

On including music from other artists (for atmosphere, decoratively, for intros/outros & transitions)

In the first instance, if you know the artist ask for their permission to use their material. You may be pleasantly surprised by the positive cycle of recognition and community surrounding artist works. Make sure you acknowledge them in your show notes.

We recommend releasing under the “Attribution-NonCommercial-ShareAlike version 4.0  creative commons licence (aka CC BY-NC-SA 4.0). You can choose from a number or variations of the CC licence; CC BY, CC BY-NC, CC SA etc.

These licences can be viewed at https://creativecommons.org/licenses/by-nc-sa/4.0

Always check the royalty free claims of audio samples found online. If you want to find audio from online sources, search for CC BY titles on Soundcloud or on YouTube

The link to the YouTube channel  "Of Musicians" below is a collection of a selection of royalty free original music.

https://www.youtube.com/channel/UCRmSat9_7JXFGAQD0V-L0-g/videos

Record the following information in your show notes to establish the provenance and bona fides of 3rd party music, artwork and creative content that you decide to use:

Music (CC BY-NC-SA licence) 

Title: “a title”

Artist: “a name”

Source: “a link to the original”

Licence: “CC BY-NC-SA”

Cover Art (CC BY-NC-SA licence)

Title: “a title”

Artist: “a name”

Source: “a link to the original”

Licence: “CC BY-NC-SA”

Example workflows:

Note, don't just apply these without testing. The basic method is to select a 10s sample and preview effects/filters until you are happy with the sound.

Simple standard process

Align tracks

Per-track noise reduction

Isolate voices

De-clicker (default) on all tracks (process run-time is typically the full track duration e.g. 15-30+ minutes)

De-esser (default) on all tracks (process run-time is a small fraction of the track duration)

Loudness Normalization (-23dB LUFS)

Compressor (default)

Makeup soft and suppress clipping on small edits with Normalize.

For an original track that had a pronounced small 'room' sound and a tendency to 'clipping zone' raw i.e. the waveform looks high, square and flat.

Align and isolate

De-clicker (default) on all tracks (process run-time is typically the full track duration e.g. 15-30+ minutes)

De-esser (default) on all tracks (process run-time is a small fraction of the track duration)

Loudness Normalization (-23dB LUFS)

Compressor (default)

Loudness Normalization (-23dB LUFS) (again)

Makeup soft and suppress clipping on small edits with Normalize.

AUMultiBandCompressor on affected track (the Sub Control preset sounded nice)

Normalize all audio to push up/down to average 0.5 waveform view (to give a typical -12dB to -6dB range with -3dB peaks)

For a track with clicks and speech that sounds very wet.

De-clicker (non-default 2 pass or more) on wet track.

Isolating second voice when two mics in the same room

Switch to Spectrogram view and select vocal patches with "Spectral edit multi tool" to fade-edit cross-talk.

AUMultiBandCompressor on affected track (the Sub Control preset sounds nice)

For a track with small room reverb and 'tinny' mic

Switch to Spectrogram view and select low frequencies 1-300Hz with "Spectral edit multi tool" to fade-edit.

Apply Filter Curve EQ - apply 100Hz Rumble filter to remove 'rumble pressure' from the audio (it feels like pressing on your ears when listening with headphones).

Apply Filter Curve EQ - apply Radio filter to remove the sensation of light 'highs' pressure when listening to the audio.

Statista as a data source...

Note, I generally do not approve of the use of Statista as a data source. Statista is a data aggregator, not a primary source. It may be useful for identifying useful original sources but I do not regard it as an original source. 

To check if UCD has campus access (Campus License International).
Click on Login and select Campus Access, you then select "University College Dublin" from the box <Select your University>. Statista then logs you in automatically.

<from Statista's website>

On Campus
When you are on campus, and within the university network, https://www.statista.com/ will be available through an automated IP-activation which requires no further logging in.
Off Campus
If you want to have access to Statista off campus, from home, or a cafe, you can use Shibboleth Access or EZ Proxy. 

https://www.statista.com/login/campus/

Use the dropdown box to check if your campus has Shibboleth Access:

DataCamp Projects

Complete the following learning goal "Introduction to DataCamp Projects".

This short introduction to DataCamp projects introduces you to Jupyter Notebooks. Notebooks are widely used and are a really handy way of playing with data, for organising and presenting research.

If you cannot find this activity, navigate via "Learn" > Projects > Introduction to DataCamp Projects (screenshot below).

Navigate to "Introduction to DataCamp Projects"

Group podcast activity

Sign-up to the "Peer Group Code of Conduct" (copies distributed in-class). You can also use this to record participation as needed.

The basics

To host a 20-40 minute podcast you will need to:

1. Identify a guest to interview for the pod.
2. Agree the scope: 6 questions.
3. Arrange a date and time for the recording.
4. Host and capture the recording.
5. Edit and finish for the pod.
6. Custom cover art, show notes (description), and announcement text.
7. Polish the finished product and publish.

Note: Cover art requirement: "Podcast feeds contain artwork that is a minimum size of 1400 x 1400 pixels and a maximum size of 3000 x 3000 pixels, 72 dpi, in JPEG or PNG format with appropriate file extensions (.jpg, .png), and in the RGB colorspace. Aim not to exceed ~300KB size file."

In this example of a podcast cover-art file I deliberately included the dotted line as part of the art.

The live podcast format

Develop your own running order / show notes. The order of questions (and indicated responses) are merely a starting point. Include links to other material you might have referenced in the show or to point to additional content. Don't forget that the show notes will include acknowledgement for the music, images, clip art etc. (creator name url copyright/license).

The show notes / running order is merely a guide rather than a tight script and the show itself always takes its own course. Avoid reading word-for-word scripted responses - unless you have amazing voice acting skills it is nearly impossible to not to sound wooden. However the activity of preparing primes and enables you to engage nimbly with the flow of a discussion.

Your show-notes can be used to share links to online versions of visual elements, references to other material etc. For the final edited podcast, assume your audience is only able to listen to the audio file.

More tips:

What does the finished product look like?

Find inspiration in some of our other episodes on the Design Talk (dot IE) podcast.

https://shows.acast.com/design-talk/episodes

Prep and activities:

Read, amend and sign the "Peer Group Code of Conduct" when your group is formed.

You should identify a guest to invite (let me know if you have ideas) but if you can't find a guest I'll help you to source one.

Arrange and coordinate the date/time and venue.

Up to two co-hosts assigned for the interview (review previous episodes to get a sense of the format).

Everyone to keep track of time, to prompt and track engagement during the conversation and to highlight comments or themes that arise on the group chat.

Each member to create their own edit, to edit the mp3 file down to a compact lively podcast with intro/outro etc. using Audacity.

Each member to create a podcast cover art version for the episode.

This class is supported by DataCamp

datacamp: clear as data

We think DataCamp is the best platform for self-paced learning for data science skills development.

DataCamp is the most intuitive learning platform for data science and analytics. Learn any time, anywhere and become an expert in R, Python, SQL, and more. DataCamp’s learn-by-doing methodology combines short expert videos and hands-on-the-keyboard exercises to help learners retain knowledge. DataCamp offers 350+ courses by expert instructors on topics such as importing data, data visualization, and machine learning. They’re constantly expanding their curriculum to keep up with the latest technology trends and to provide the best learning experience for all skill levels. Join over 6 million learners around the world and close your skills gap.

The value of using DataCamp

The DataCamp skills self-paced training is directly relevant to digital and business in general. Employers recognise and highly value even basic python, R, Excel or general data science skills. Everyone's paths will be different and DataCamp supports different learning needs, offering support and challenges suited to both beginner and experienced learners. More importantly the data analytical tools are directly relevant to your term paper project, whether you use python, R or Excel, ultimately the DataCamp courses will help. Some projects in DataCamp may align well with your own research projects.

An introductory pathway for new starters and programming rebooters.

• The Excel Fundamentals track
• The R Programming Fundamentals track
• and many others.

Consider following one of the many guided or unguided DataCamp projects by browsing through the catalogue. For example:

• https://learn.datacamp.com/projects

A relevant example (using R) is "Visualizing Inequalities in Life Expectancy"&nbsp;

• https://learn.datacamp.com/projects/166

Some thoughts on learning using DataCamp

Taking the DataCamp learning tracks early and taking them seriously is important. The knowledge and learning you will gain will stand to you in your career, although you may never end up coding for work you will have acquired a deep appreciation for the work itself and an understanding of the concerns of the people who do and who you will interact with or manage.

"You will inevitably encounter specific learning challenges in different tracks. The programming courses (like computer programming in general) will pose challenges of syntax, logic, even spelling. Yes, coding can be extremely frustrating for us all, regardless of experience. That is why we also offer in-person tutorials. We believe that self-directed learning still needs an element of interaction with others to succeed, to help problem solve, to overcome roadblocks. We have all encountered education models where information is presented pre-configured, where the answers are provided in plain view, merely requiring extraction and mapping to solve problems, or regurgitating at the right moment of a multiple choice question. But if you find yourself stuck on a problem and unable to make progress based on the instructions. What then? The educators in DataCamp, while providing lots of scaffolding and spoon-fed problems, actually encourage, and in many cases require, the learner to go off and explore other related content and solve problems using external resources. Using "search", drawing upon and participating in online communities such as StackExchange and other sources is both appropriate and commended. Not for copy-and-paste-code, but for phrasing the problem, following dialogue, engaging in conversation. Over time you will find yourself emboldened to take an active part in these communities, supporting others as they struggle problems like those you eventually overcame."

Case Based Learning CRIB Sheet

A process view for organising case based learning (aka problem based learning); to formulate self-defined and self-directed learning goals.  Cases can be analysed at progressively deeper levels, from simple to more complex and/or subtle analysis as you become adept at this approach to learning.

With respect to the cases, you will get out as much as you put in. Consider using the following process and prompts.

• Case analysis process guide:
• Read the case
• Identify what you already know about a topic. 
• Develop a list of potential problems evident in the case.
• State learning gaps then resolve them by finding and reading-up relevant material.
• Summarise your findings and ideas.
• Generate a synthesis.
• Present a single page capturing the above (homework).
• Retrospective (reflect on the process).
• Case analysis self-question prompts:
• Are there topics/objects in the case you need to know more about?
• Do you have applicable prior learning and experience you can bring in?
• Did the case raise questions for you? How did you answer the questions?
• Is information presented that you do not understand? Avoid highlighting non-specific generalised experience/skill gaps that you cannot address through independent research and learning.
• How did you resolve your personal knowledge gap(s)?
• Our intention is for you to show that you started to address the knowledge gap here. So, what did you learn?  
• Any general benefit? Is the case applicable to inform future practice?
• Are your recommendations, prescriptions, statements or claims justifiable?