Watched over by Sir Geraint Evans, good cop bad cop begin their WalesLab research project at the Royal Welsh College of Music and Drama in Cardiff.

Unaccustomed as we are to public blogging we offer the following notes on the days’ work: (please note that less note taking occurs during periods of greater immersion in, and interest in, the task in hand - we'll leave you to work it out)

Text to speech day 1

Coffee and catch up

The room!!

Timing. - Blank line adds time over a full stop, extra blank lines add nothing. Same for colons and semi colons.

John learns that in writing there are supposed to be spaces after full stops.

I suppose what we are trying to do is to put space into the words. As with any content the space is part of the content.

A line break between words is unlistenable!

Recording out and back in to Logic Pro to analyse waveforms in order to detect effects of various punctuation - only question marks seem to have slight effect.

Wandering off into logic pro, double tracking and other twiddling. Wondering whether possible, and desirable, live? Could probably be patched into ‘Ableton live’ if wanted, but probably a bit too tricksy for our needs?

John discovers short sentences are best

Playing with words it cant say

Preference for Vicki at half way between slow and normal

Auditioning the potential 'cast'

Alex - new voice for leopard - laid back Californian, not bad (see below)

Bruce - hmmm

Fred - Stephen Hawkins (when slightly slowed)

Junior - sh*t

Ralph - Andr*w M*l*n

Agnes - Quite bossy

Cathy - bit asthmatic

Princess - simple - blow up doll

Vicki - our favourite - pussy galore, Diana doors

Victoria - hopped up

Albert - jack Hawkins, only to be used while smoking

Bad news - funeral march

Bahh - fairy tale dwarf, needs to clear his throat

Bells - pretty much useless

Boing - Cute little jack Hawkins sidekick robot smoking a bong

Bubble - talking while blowing into a bong

Cellos - hall of the mountain king

Deranged - Possible stalker for Vicki

Good news - Land of hope and glory - comic effect only - no speed control

Hysterical - unintelligible/idiotic

Pipe organ - good for single words

Trinoids - useful aliens

Whisper - man speaking from inside a freezer

Zarvox - Good upbeat robot

Sidetracked by the asthmatic tendencies of Alex - explorations, trimming punctuation, words, and letters in order to generate the breath intake. But it doesn’t seem to be based on word count…inconclusive, possibly, but maybe too time consuming.

Lunch

After lunch

Starlust - straight from Preview pdf, text scanned as a pdf read through preview, imperfectly but interestingly!

Wondering if we have already spent too long listening to the artificial voices as it appears to both John and me that Vicki is actually reading rather than just ‘voicing’ or speaking.

More research needed on readability of ‘texts’ in pdfs. When the text tool option is available then it would appear that even hand written marks are readable.

Pobf !

Something to do with original scan settings probably? Will need to check this through making original scans.

How about if we get ‘speech to text’ software to speak it back?

My favourite line of the day, ‘pants and a pink negligee’

Alex at his slowest is difficult to beat!

Figures up to 999999 are spoken as a single number (i.e. nine hundred and ninety thousand, nine hundred and ninety nine) after that it speaks each digit individually.

End of day one (RHM)