Assignment 1 Feedback

During the presentations on tuesday, each class member was required to write out feedback and responses for each presentation. Ok, so only 16 people actually handed them in. And ok, so one of those was mine. And there were quite a few that just said "good presentation". But I went through and cut out the ones that were relatively interesting, and here they are :

Most of them I can read. That bottom one I've no idea, it had squids and stuff drawn all over the sheet. Someone must really love their cuttlefish.

Most of these ideas I reckon we should take on board. Roughly, I think this is the translation :

1 - I had reservations on user answers, and think these be manually screened
2 - Is this really suitable for an art gallery? more likely to suit a shipping center( re :P )
3 - Voice Activation / database to ... something
4 - Will the person hear the computer fan ? How willing are people going to be to follow the rules?
5 - Generic comments
6 - Research other projects which are similar
7 - Excellent presentation even though the idea is not too convincing ( maybe the idea just doesnt interest me )
8 - Chance that people will say something things? Limit it to horoscopes to get more generalised response. Measure cost, dont trust megan ( get broke? ) The recieving box doesnt need to use voices to choose

That's as best as I could decipher them. . . They say the more creative you are, the messier your handwriting , we must have a very creative class ;)

On top of these responses, during the presentation, a few people had suggestions.

• Research the usenet oracle ( http://cgi.cs.indiana.edu/~oracle/index.cgi )
• Limit / Eliminate voice recognition
• Use Physical Answer System ( eg, tarot cards, crystal ball )
• User answer screening ( "was this helpful ? yes, no " )
• Manual screening of answers each night
• Two shadows - one for question, one for answer ( two voices )

I think that's everything. Pierre, megan did I miss anything ?

So guys, comments / ideas as to where to take it from here ?

Hey...

Sorry about missing the presentation, hope it went alright - been really run down lately, was at the doctor's yesterday and had a blood test today to see what's going on. Feel so ill now :( got a medical certificate and all that

Once again sorry about missing Tuesday - can anyone fill me in on the next step for the project?

Email about report

Below is a copy of the email is sent to comp3000 regarding our report.

"Hi,
This email is on behalf of our group. We found out today that there was a mix up on the weekend and 2 sections were not included in the report. We have a blog in which we collate all our thoughts and work. I do not have the internet at home so we agreed that I would post my section of the report on our blog by Thursday night. For some reason, that was not then transferred into the report even though other sections that were on the blog were. We don't know if it were a technical error or not, but we would appreciate this to be taken into consideration on the marking of our report as all sections were completed.
Thanks
Megan (on behalf of Silent Radios)"

Tomorrow's Presentation

Ok, Well im thinking that because noone contacted me that you didn't need me to give a hand with anything. Just wondering what we have planned for tomorrow? If anything at all. We didn't talk about that in the meeting last week from memory, so if you want me to prepare anything, please txt or call me. I don't remember what we had to do.
Cheers

My Contribution - Background Research

Hope this works for you guys : Copy and paste follows

Survey of relevant background material

Concepts and Inspiration

There are many underlying concepts that run beneath this project, each of them is as important as the next. When the group was given back the suggested ideas at the start of the semester, we were presented with two distinct directions for development. One path lead towards conversation monitoring, sound manipulation, speech recognition and sound analysis – While the other, lead towards life-as-art, responsive seating, reflection and sound feedback. It was our job to join these two distinct areas together into a cohesive new project.

Eventually we came to the idea which is being presented in this report, an “oracle” that gives advice, based on a certain keyword related questions. The base concept that we are using as a guide would be an “oracle”. Oracles traditionally were those who could predict the future. They would often offer advice on a certain situation and give guidance and council for those who were searching for it. Of course, this is the ancient traditional oracle, if you were to examine the equivalent today, you'd be reading horoscopes, calling psychic hot lines, or getting a tarot card reading in a shopping centre mall. This is the foundation for the physical aspect of our project. The technical and aesthetic aspects were inspired by a combination of elements including each group members' specialities and interests.

Related Projects

“The Turing Test” and Artificial Intelligence
While this test is not specifically a related “project” the concepts behind this test are closely linked to the ideals behind our project. The Turing Test ( Designed by Professor Alan Turing ), is a test to see if a computer can mimic human conversation. In short, the test requires that a human judge has two conversations, one with another human, and the other with a computer. If the judge can not tell which is the machine, and which is the human then the machine passes the test. The Turing Test is a test of artificial intelligence, and in a way, if a machine can pass the Turing test, then that computer would almost appear to be a human being itself. With our project attempting to simulate the presence of another person, some form of AI, or AI behaviour would help to hide the machine behind the exhibit.

Talk2Me, at ReActive exhibition – Ann Morrison
A related exhibit which a previous studio lecturer ( Ann Morrison ) put together at “The Block” ( QUT, Kelvin Grove ) Involved a user speaking into a microphone and “talking” through a computer. Although the exhibit only had a small amount of text to speech conversion, there was still enough interactivity there to show the possibility of future implementations. A brief email from Ann mentioned that she used the inbuilt mac osx speech recognition engine to get the user input
http://www.itee.uq.edu.au/~morrison/ for more information visit http://anmore.com.au/talk2me/talk.php

AuraLamp
AuraLamp is an experiment in “contextual” Speech Recognition. Ie, and object that knows when you're speaking to it. Currently with speech recognition software, it always assumes you are talking to it ( usually controlled by a “stop listening”, “start listening” command ). This project tracks the Eye Contact of the participants, and only responds when someone is looking directly at the lamp itself. Our project considered a similar problem of knowing when to listen, and when to ignore. Although the eye contact technique wont work in our chosen environment ( instead, we've gone with the pressure sensor , or similar system ), the concepts behind this project aided us in our realisation.
http://ubicomp.org/ubicomp2003/program.html?show=demos Title - “AuraLamp: Contextual Speech Recognition in an Eye Contact Sensing Light Appliance”

Technologies

The main technologies used in this project will be speech recognition, as well as the ability to analyse, and store recorded messages depending on a category or keyword.

Speech Recognition is an extremely complicated and ever-changing technology. There are many strategies for implementing it, as well as many algorithms and techniques for acquiring the speech data and making it available for analysing. Speech Recognition is the process of converting a speech signal to a sequence of words. For example, someone says the words “I like studio” into a microphone, the computer recognises the sounds produced and the words they make, and can reconstitute text or commands based on that information. Modern day speech recognition systems use the “Noisy Channel Formulation”, in which the system will search for the most likely word given an incoming signal.

Our project will rely heavily on speech recognition as the users will be speaking to the exhibit directly. It must either be able to store sound data as a recorded file, or analyse the incoming data, and store that information as text.

Further technological requirements would include a database system for storing information, as well as a way to provide possible statistics for users as they interact with the system.

A variety of options are available as to a technological source for the majority of the programming. Currently the best option seems to use the program “Processing”, which allows for dynamic recording of speech, as well as possible modification and analysis. Other options include using external programs for individual elements with a core module tying them all together.

Sources

http://www.turing.org.uk/turing/
http://www.turing.org.uk/turing/scrapbook/test.html

http://processing.org/

http://www.ling.lu.se/research/speechtutorial/tutorial.html

Soooo...

So guys!
About that project that's due tommorrow ?

Constraints

some quick dot points of where I was going with the constraints

ambient recordings + difficulty separating peoples voices (too much ambient noise) = developed into 1 on 1 private booth concept

Wide range of possible questions -> limit this by selecting a set of keywords

People abusing the system / giving stupid / nonsensical answers to computer’s questions -> perhaps a screening process at the end of each day?

People giving abusive or offensive answers -> introduce a filter that recognizes common offensive material in the speech pattern and ignore such responses

Background noise and interference to mic -> try to eliminate or reduce this by using heavy curtains or other soundproofing methods

Confined space of the art gallery -> potential crowds in a confined space = heavy background noise – furthers the need for soundproofing

Storage space -> depending on the length and quantity of responses, audio files could quickly fill up whatever storage medium we decide on

just an outline
any suggestions or changes?
mitch

anthony...

Does the whole report need to be paragraph? I wouldn't think it would be absolutly necessary so i left some parts a bullet points. To me is says more and is easier to read. Is there anything else you need me to cover or fix or whatever? If there is txt or call me.
Cheers
Megan

Contributions/Responsibilities

Anthony - Unofficial group leader. Coding and technology that is behind the artwork, note taker.

Pierre - Construction and visual aspect

Megan - Graphic artwork, audio, visual aspect

Mitchel - Graphic artwork, construction

This is an early outline of what the group members contributions and responsibilities will be over the semester. We will all contribute to our group blog in order to keep up to date with where we are at. This is not cemented as there will be circumstances where we will need to contribute to other areas.

Target Audience/Experience

Our target audince:
- generally art gallery patrons
- those that are superstitious and interested in horoscopes and astrology
- younger but mature generation
- curious sorts
- we will probably attract a more female rather than male audience

Visual Experience:
We want a more traditional look about our interactive artwork compared to a "Miss Cleo" view. What we mean by this is that we would rather have cushions, tapestry insense and oil burners, more voodoo rather than the more teenage crystal balls, tacky view of the experience. It is more or less a level of maturity that we want to engage with.

General Experince:
We are likely to get people who will abuse the artwork and not take it seriously, we can't take this away from the overall experience so we will incorporate that into it and accept it. We want it to give people an overall uplifting feeling, and to walk away happy and impressed.

Keywords

A list of keywords for responses
Love
Wealth / money / fortune
Health
Happiness
Relationship
Fulfilment
Luck

Also I need to find a scanner or a drawing tablet so I can put my sketches up :S

- Mitch

8 Keywords

Keywords that can be used to trigger particular responses:
1. Love
2. Happiness
3. Wealth/Rich/Money
4. Future
5. Relationship
6. Emotions (anger, sadness...)
7. Children
8. Health

Megan's Sketch

Here is my interpretation of our project. I do believe though that we aren't all that unique with the whole fortune booth idea... Do we think that will matter?

To Do List.

For the report, we've initially broken it up into the following sections :

1) Background Research - Anthony
2) Target Audience - Megan
3) Constraints - Mitchell
4) Timeline / Milestones - Pierre
5) Contributions and Responsibilities - Megan

That's taking the sections ( in order ) right from the criteria sheet. I havn't looked at the template yet, cause it seems to be a .dot file and its over 10 mb!? What the!? I'll try it again later.

okay, everyone by tommorrow, SKETCHES of what you think the project should look / feel / work like, and at least 7 possible keywords. Post them here to the blog.

Meeting #3 : Final Project Idea

So we sat down and finalised the project today, notes are as follows :

Interaction :

1) the user enters the "oracle" chamber / room
2) the "oracle" asks the user a series of questions ( Age, Star Sign, Sex, And then a series of keyword questions ).
3) The user can then request information / proverbs / advice based on one of the selection of keywords.
4) The oracle searches it's database and provides a response. Either TTS or Recording of the person who originally said it.
5) Person leaves feeling wonderful.

Where does the oracle get its advice from?
Gets its advice from the "Question Exchange". In order to ask a question, you must first answer a question.

Possible Extensions :
- Dynamic Keyword list
- Basic AI, Understanding base conversations

Physical Representation :
- Booth
- Oracle / Tarot / Fortune
- Enclosure to dampen outside sound sources
- Curtain / gause / with Silhoutte of a figure behind the curtain.
- Pressure sensitive seat / coushins

Did I miss anything guys ?

Ann's Response

I sent Ann an email last week, asking her how she accomplished her speech recognition in her dome project that was displayed at "The Block" at QUT. She replied the other day, stating that she used the mac os x speech recognition, and she merely re-trained it ( in short ).

I think for our project, we'll require something more complicated, but at least that possibility is there.

Chinese Proverb Breakdown

Ok, so we've been mega slack with our timeline here. I'm not even sure if we've got a final idea decided upon yet.

From what I gathered at the last meeting, everyone was pretty happy with the “chinese proverb” thinggy that was talked about. So I've sat down and tried to map it out a little better. From what I could see, there were two ways for us to approach this. I've broken both possibilties into what I call "Oracle 1" and "Oracle 2". Before I get to that though, I looked at breaking the idea down a little more.

Chinese Proverb Oracle -

What does it do ?

Gives advice based on key words spoken by participants

Where does it get the advice from ?

Manual Prompts / stored data, or ambient recordings

How does it display the advice ?

Text to Speech, or Combination of pre-recorded vocal samples, or a combination of the two

That's what I felt was the "Oracle" in its simplest form. Going from that, I thought it could be split into two parts. "Oracle 1" , and "Oracle 2".


Oracle 1 -
Provides inspiration advice based on ambient recordings from around the space

REQUIRES:
- Voice Recognition
- Analysing of vocal data
- More Specifically, Understanding what has been said
- Vocal Dictionary
- Dynamic Sentence Construction.

Oracle 2 -
Inspiration advice provided via specific user interaction

REQUIRES:
- Storing Answers to Specific Quesions ( Keywords )
- Voice ( Keyword ) Recognition
- Dynamic Sentence Construction.
- Question / Answer Storage


From this point, I thought about Oracles throughout history, and the tradition / mystery behind them. They predict the future - good or bad. Though the predictions are usually vague. And that made me think of those ancient chinese proverbs you here all the time. such as

"An ant may well destroy a whole dam"

or

"Crows everywhere are equally black"

Incredibly cryptic stuff like that. And that got me thinking of those automatic cryptic proverb generators you see on the net these days. It was at this point I got myself all confused and decided to break the whole project down again.

I drew a diagram, but my scanner is dead, and the drawing is rubbish - so I'll just do my best to describe it here. I thought, what is the most fundamental element of this project, it Gives Advice. So then I thought, how does it know what advice to give? it's Based on a keyword or phrase. Which makes sense. Then I had to think where these keywords / phrases were stored, and taken apart. There's a possible two options. They're Randomly Aquired and Generated, or they're Pre-Defined Responses ( user submitted ). The Randomly Aquired elements would be taken from Microphones submitted around the environment. but how to distinguish what's what? Does Placement Dictate the data that is being recorded? or will it purely be a collection of words. In which case, does it need to understand the context of the words at all ? or merely just the words themselves. As for the pre-defined responses, we have to wonder how we can garuntee that users will submit responses. I considered some sort of Offering system. You Offer an answer to a question, or a word to the "oracle" and it will then grant you an answer to your question.

From there, I had to consider *how* this information would be presented to the user. Via Proverbs seemed like a good idea. Randomly generated proverbs based on the recorded vocabulary of the system. Though then, would they be Played Back as the actual recorded voices? or would they use a Text to Speech ( TTS ) system, or Both?

And then, what about a visual response as well ? A wishing well, with the proverb written beneath the surface ? or merely auditory. A hole in the wall, a whisper on the wind, multiple voices all replying at once. The three fates giving a little bit of a hint to the questioning mortal.

And that's what I sat down and thought about.

Group, I need your feedback. We're way behind on this one, so yes - we've got to get all this sorted tonight, but more likely, tommorrow. So we can get this assignment written, and organised.

Sorry if this was just a big long rant, it's taken from a bunch of a4 pages of notes I threw together roughly. But it seems to shed a bit of light on the situation, at least, on what I feel the project is about. What do you people think ?

- Anthony

Timeline

At our group meeting today, we talked about a draft timetable for the proposal assignment. The proposal is due on the monday the 26th ( presentation on the 27th ). So we've come up with this draft timetable :

Wednesday, 14th -

• Have all ideas and concept blogged by wednesday night

Thursday, 15th -

• Thurs morning, examine the ideas on the blog, decide on the final project
• Thurs night, have final project idea chosen

Saturday / Sunday ( 17th, 18th ) -

• Allocate assignment parts

Friday, 23rd -

• Have proposal completed

How does that sound for everyone ?

Email to comp3000@...

Here is an extract I emailed to the comp3000 address:

"Our group would like to do some interactive research similar to what was presented in the lecture today with the whiteboards in the library. We would like to record people's opinions, reactions or answers to questions on certain topics or emotions. We would like to leave a tape recorder and a microphone at the entrance to the library to do this, or Di's office (preferably the library as there is a greater traffic flow). Eg. We pose a question: What do you say to someone who says they feel angry? The person records their response into the tape recorder. In the long run we are looking at having a collection of descriptive responses to different key words that can be played back to people if they use any of the key words in talking about a particular artwork in a gallery."

Do you think that explains what we want to get across? It was kind of hard to describe. I also suggested if more info is needed to come here. So if any group members want to contribute, change or fix what we need to, please do.

-Megan

Meeting #2 : Notes

We got together with our new group member Mitchell today, and we sat down and talked about a few more ideas. The notes I took are as follows, details / timeline / to do list will be in a post I'll write later this afternoon. Sorry if they're incoherant, I dont have the concentration to re-write them at the moment, they're doing construction work next door. Jackhammers are very...very distracting.

Proposal idea -

• See if people will actually record willingly.

  • Leave tape recorder in a public place and see what people say.

Figure out a way to divide assignment into even parts. 25% code ? 75% ... ?

Direction – recording thoughts, reflection.

Reflective point, seating ?

Listening point for other objects.

IDEA LIST :

• Wall
  • Disguised speaker
• Chair
  • Sit down and given sound.
  • Sensor in the seat, plays sounds
  • Sound is ambient, unknown sound source
  • The empty chair
  • Multiple chairs for multiple artworks

• Enclosed area
  • Confessional
  • Surrounded by sound, sound is part of the environment

• Artwork on the wall
  • Seat in front of it as well
  • Mini lounge

• Collaborative artworks

  • Collaboration
  • plus possibility for expansion

• “Chinese proverb oracle”

  • Gives advice
  • Recording advice
  • Inspiration.
  • Keyword assosciation
  • Ego boost
  • comfort

CVB :

• Random noises and sound effects. And Sighs.
• Word association based on the emotion the artist feels their artwork / piece displays.
• Single word association.
• User Submitted responses.
  • Using tape recorder

  • “What would you say to someone who's sad?”
• Recordy thing.

Mitchell ?

I just had a look at the blog listing on the course website...
Have we aquired a new group member ?

According to the list we've got Mitchell G. Chapman in our group too ? Am I missing something?

-Anthony

Furthermore, why doesn't Megan have a middle name ? :P

Meeting #1 : Notes

Meant to put these up ages ago - but I've had a silly / busy week. Various notes from our first meeting as a group. Exploration of ideas and possible constraints from the initial suggestions and initial 3 ideas.

“Silent Radios” :
Anthony Massingham
Pierre Medeiros
Megan Daley

Our Ideas :

Megan : “Toilets for Seats” - Toilet seats with sound effects

• Various Cost Issues
• Used for Reflection on life / artwork
• Novelty
• Abstract idea

Pierre : “unsure of name” - Recorded Conversational snippets

• Privacy Issues

Anthony : “Chatter Critic” - Recorded and Analysed conversations

• Privacy Issues

General Comments :

Massive ethical issues with all three ideas. Privacy with the microphone / recording ideas, Costs with the toilets.

How these ideas link :

• Recorded Sounds
• Sitting and Reflecting
• Listening to other conversation/ideas
• Expansion on ideas/thoughts.

Expansion on ideas :

On Recording -

• Ambient or manually recorded and saved in a record bank.

• Randomly Played Recording.

Possible Implementations -

• “The Walls have ears” - Physical representation, an actual wall where people sit and listen to what others have said.

• Speakers spread throughout the space, filling it with general ambient noise / conversations

• Actual physical art prescence

• Confessional / closet

  • Similar in style to http://grouphug.us/ ( Anonymous Confessional forum ).

    • Would require some sort of voice synthesis for anonymitiy.

• Speech analysis. Breaking conversations up into individual pieces. Possible ? How hard ?

• Lost Voices. Excerpts of thoughts. ( Lost Souls, Bus Stations . . . Disembodied voices... Ghosts ?)

Our 3 ideas

Here are the 3 ideas that our group have started with:

1. Chatter Critic ( Anthony Massingham ) -
An ambient and background conversation monitoring tool. Microphones are placed throughout the gallery at various artworks. Each microphone monitors the conversations that people have about each painting / exhibit. Each conversation is analysed for positive and negative comments. These words / phrases are then sent to the computer, and are saved, edited, cut apart and re-arranged into new phrases and comments and played back through speakers as well as being displayed on screens. Depending on the frequency of use of a particular word of phrase it will determine the amount of times it is repeated or displayed.

2. Toilets for Seats (Megan Daley) -
Art Galleries need seating, particularly in front of artworks on occassions. So, an abstract interactive artwork can be made involving toilets as seats. Here is a scenario: Person walks up to artwork on wall and admires it. They turn around and see 3 toilets in a row, all artistically painted and clearly shown to be seats for admiring the artwork on the wall. Person sits on toiltet seat. After short period of time, sounds of someone going to the loo can be heard, this soundtrack continues until the person gets up and walks away from the toilet and artwork on wall. After this a flushing sound is instigated. This can break the some-what serious nature of many art galleries.This would work with sensors on the toilet seats and speakers inside to create the effect that they are using the toilet.

3. Pierre's is to come...

Intro to our Studio 5 group

Hi!

This is our group blog for Studio 5 - comp3000. Our group is called Silent Radio (Whispering Closet), and members are Megan, Anthony and Pierre.

This blog is a site for discussion about ideas, research and revelations. It is contributed by members of our group with posts from all 3 of us frequently (hopefully). It is open for feedback from anyone that would like to give it.

Cheers