Speech Reviews

Reviewers this week were Luke Dubois (LD) and Paolo Pedercini.

aahdee PP:
Nice idea and execution, voice control has an inherent relationship with authority and its impreciseness works with the kid and delivers the lazy dad satire
LD:
i like the story, and the way in which you keep reminding the user that they’re talking to a toddler. so i would add some feedback to the system so you know that it knows that you did the right thing, e.g. have the text include the adjectives you use (‘big blue’ but also ‘small yellow’ if you say it).
GL:
I don’t understand the documentation. Is the background speech somehow part of the project??
avatar PP:
I wasn’t familiar with the toss my salad idiom and I don’t feel smarter now that I know about it. The sketch is ok, it’s not clear who is tossing and being tossed and how the visuals and voice activation support the innuendo
LD:
i’d build it out a lot more… what if you say something else? if there’s only one phrase that does anything at all maybe make it about how you say the phrase and have the video change behavior accordingly.
GL:
It’s not clear that you took the opportunity of the assignment to think deeply about what it could mean to have a conversation with a machine. As an interaction, it is simple to such a point that it begs the question as to why not just have a button to toss the salad. I suppose it is transgressive for the user to state the phrase out loud. Anyway, the project functions as advertised: the user states the phrase, the machine responds with an animation.
conye PP:
Nice, the withering is heartbreaking and the ryme bonus could make into a nice language game. The background pattern is not visually consistent
LD:
a lot of variety and agency on the part of the speaker to play and experiment with what kinds of works are good and bad for the plant. i love that rhyming gets you extra points. good UI design showing user voice feedback at the bottom.
GL:
Issuing commands to a flower is oddly gratifying. I appreciate that you were able to (technically) detect rhymes, but it’s not clear why rhyming should “double the impact” of commands. This feature is also not self-revealing; it’s not clear that this is something one could or should do. The idea of a software game/artwork that elicits rhymes from the user is interesting, it’s not clear that this is the right concept/metaphor for this interaction.
creatyde PP:
Phone consciousness / urban legend robocall is an interesting idea but the execution is lacking, I’d look for way to encourage interactivity and make it spread virally from phone to phone as a kind of prank or participative project
LD:
i like the idea and i like the result, but the interacting technology might be overkill for a piece of generative sound art. you might want to mock this up as a web site first and see how you can get a variety of responses.
GL:
This is a seriously interesting and intrepid investigation. But the documentation leaves much to be desired, it’s almost impossible to understand what’s going on.
dechoes PP:
The alliterated answers are really beautiful but I’m not sure about the grandpa character, dementia is a sad condition and the family setting may turn a nice digital oracle into a joke
LD:
so this is a great implementation but maybe not the best concept… my mother died of alzheimers and would go on reveries like this for hours, but the ‘gaze’ here is making fun of people like that, not creating anything cathartic or enlightening; so i’d tweak the ‘plot’ to find a less fraught subject to poke fun at.
GL:
Well-done. But is it funny in a way that punches down? Worth considering an interlocutor (an alien blob?) that doesn’t lampoon people with dementia.
farcar PP:
Nice, I’m not sure about the scan and metre, if you are using shakespeare I expect formally impeccable verses. It should stop after a rhyme and request a new input to make it a really collaborative endeavor
LD:
this is totally cool. it would be not super-hard to ‘upgrade’ it to understand a more complex metric form that was used by shakespeare (e.g. iambic pentameter, like the pentametron).
GL:
The documentation is too short; the results are difficult to evaluate. But this seems very well done.
jackalope PP:
Witty, the parrot should talk
LD:
i love the parrot but why doesn’t it talk, e.g. using p5.Speech() with crazy speech synthesis settings? i think that would be a huge bonus to the project, to hear an actual cursing profane parrot.
GL:
Well done. Documentation is a little confusing; should we be hearing the parrot? The text is well synthesized.
joxin PP:
I couldn’t make it work on my browser but the idea of an interactive podcast that checks with you and acknowledges your emotional response is super intriguing. I can see it potentially developing into an experimental journalistic project
LD:
this looks like a super cool project but i couldn’t get it working… you should have a video render ready of the piece. i looked at the code and it wasn’t immediately obvious what was wrong, but something must be missing.
GL:
Very well-conceived investigation/prototype. Documentation would help illustrate the interaction better.
kerjos PP:
LD:
this is great, and could easily be multiplayer, or with all sorts of other bells and whistles to make it a crazier game (like the number of bots grow). you could also add speech synthesis to it to make it really feel like a chorus of bots.
GL:
Terrific investigation!
miyehn PP:
Very rudimentary but as a proof of concept is really promising, I’d love to see a voice controlled cosmic god/king of the universe game
LD:
terrific idea, and realy clear interface for playing with natural language… i would add visual text feedback from the speech recognizer (maybe at the bottom) and some instructions on the site. but it is really cool.
GL:
Great work. Strong, aesthetic implementation.
ookey PP:
Association chains are always charming but a bit thin and the speech doesn’t add much
LD:
this is really good… simple and clear, and i love the speech synthesis feedback.
GL: Nice work!
phiaq PP:
That’s so cruel and funny. Refreshing
LD:
this works fine, but it might not be a great idea to encourage people to work out their shit by insulting someone, even a fictional 3D head, and having it grow pimples in return.
GL:
This could be seen in different ways. As an expression that manifests (say) the artist’s own anxiety about their appearance, making a project like this could be a good way of getting stuff off one’s chest. On the other hand, this application could be seen as a tool to reinforce this sort of (verbal) abuse in the world. In such a case, it punches down; I’m not sure I can condone that.
rolerman PP:
Fantastic, I can see this developing into a more polished project narrowing it to more specific literary genre(s) and adding variables like a randomly generated cast of characters to ensure consistency and predefined narrative structures, check Vladimir Propp’s research and Once Upon a Time, the card game
LD:
great project, really good documentation, really good sample of the kind of output it can have.
GL:
WHOA, hells yeah, great concept
sheep PP:
Very cool, I don’t know if the voice is 100% justified since it’s not very expressive, perhaps it can be a screen less game to be played while taking a walk or driving or an accessiblity plugin for text adventures
LD:
i love the zork feel to it, upgraded for speech recognition and synthesis. non-oculocentric design for games is hard to come by. you could add classic text adventure things like inventory next, so you can hold things, use things, drop things, etc.
GL:
Very nicely done.
tesh PP:
Interesting concept but the text generation seems very limited. It would be cool to find a way to plug it to an actual game server and give voice to npc/bots by triggering generated speech according to in game events
LD:
this is wonderful and funny and absurd. i feel like the friends should respond a bit faster and maybe get in arguments with one another as well.
GL:
Good experiment.
tyvan PP:
sorry it didn’t work on my browser / I couldn’t try it properly
LD:
doesn’t seem to work very well… it hijacks your browser by not giving any way to stop the speech synthesis of the wikipedia page. also the wikipedia image is offset and doesn’t scroll. i think the ‘ask a question’ for wikipedia thing is great, but the result maybe should be a script that tells the browser (not openprocessing) to open the right website and then read it using the accessibility features.
GL:
Documentation is missing; part of this course is concerned with explaining your work through documentation. Examining your code to find out what you did, the idea doesn’t stray far from the provided template.
zaport PP:
Incomplete and not much different from a standard query
LD:
shouldn’t be too hard to get a working demo together once you fix the code.
GL:
A good start and laudable ambition. I wonder if Twitter is the right place to find the kind of stories you’re looking for.
zbeok PP:
The end result as performance is quite funny and complex considering how simple the algorhitm is
LD:
this is very close to being awesome… it has the right tone of frustration and melancholy, but needs just a touch more contextual tweaking in the language setup and the way it listens to you. it’s robotic but needs to be a little bit more up the curve towards the uncanny valley.
GL:
There’s a lot of good development here, better than you realize.