| |
| |
| |
|
Page: 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18
Comments:
<0> having said that
<1> it sounds (pun intended) like a fun projects
<0> my friend is doing automated translation from japanese to english
<0> I wonder how well he got on
<2> speach to text is hard enough as it is
<0> in a few months time
<2> with music through it is just too much
<3> fermat's last thoeorem is done, but only few people can understand the maths behind it
<1> Kalasanova speech to text is a different problem (wildly) than recognizing the voice and removing it from a signal (speech to text IMO is a more difficult problem)
<0> I agree with vawjr
<2> youre right about that
<2> but still
<0> no semantics
<1> though there is a helluva lot of redundancy in speech
<1> well, in language
<2> only if you talk very clearly the computer understands you
<2> singing is far from it
<1> Kalasanova the information content in written English is quite low... iirc around 1 bit per character
<2> how on earth would it take screaming and guitar sound apart?
<1> Kalasanova frequency signatures, I would guess
<2> have you ever listend carefully to guitar sounds?
<2> to the distortion?
<1> along w/ envelopes
<0> then there is the question of any generic voice
<1> one's presuming that the voice isn't doing a Wagnarian opera
<0> or just one particular performers voice
<1> i.e. there's actually some melody
<2> but many sounds cover a large area of the spectrum
<1> it's not clear how much the human voice does
<1> we were commenting earlier that we've not seen references on it
<2> filtering speach out of it isnt the biggest problem, but keeping out some of the music is the hardest bit i persume
<1> I think the idea is to leave the music "untouched"
<3> it is, but teacher said he would tolerate minor errors
<0> lol
<3> like try to make it as less as possible
<1> "minor" that's real nice of him
<0> your teacher is insane
<1> Quantum`` did you see the "1 month" for PoC
<0> PoC?
<1> Proof of Concept
<0> LOL
<2> i dont agree Quantum``, he just doesnt know anything about programming and sound
<0> ah
<0> then his teacher is ignorant
<2> yup
<2> totaly
<3> what is proof of concept?
<1> well, use either a Sade (is that how she spells it?) or Johnny Cash song
<3> like making it work on paper?
<1> they both have a range of only about 5 notes
<1> which should hlep a lot
<2> getting an other teacher might help more
<2> one that has more understanding of what is feasible
<1> I think I'd almost rather write a program that "listens" to the music and writes MIDI file to 'play" it
<1> though listening to violins get's complex.... from what I remember the main tonal characteristic of the undamped high string on a violin is the 13th harmonic
<1> and all 4 strings have different tonal signatures from each other and when damped or undamped
<1> and that's not a voice making different vowels/consonants along the way
<3> i don't know what to say
<3> do you think i could make idea in next 2 weeks?
<3> netsplit?
<1> neosisani depends on what you mean by "make idea"
<3> well projects are done on 4 seminars
<3> on first seminar you get a bunch of lectures on different topic and are given a project
<3> then you have one month to make idea, after which (if you are invited) you get another week for theoretical development of idea
<3> then you have couple of months to finish theory and go again to code program
<1> I don't know what "make idea" means
<3> well it is like to make simple algorithm, but not into details
<4> can anyone help me with a multiple decleration error? the source code is at this location http://www.noidea128.org/sourcefiles/15979.html the error is the following
<3> like find algorithms you need
<4> /home/nush/Matcher/old2/QuadraticProbing.cpp:19: multiple definition of `isPrime(int)'
<4> /tmp/ccVBDjEp.o:/home/nush/Matcher/old2/QuadraticProbing.cpp:19: first defined here
<4> Node.o: In function `vector':
<5> CHAR_BITS varies per platform correct?
<4> thanks guys
<6> Cowmoo: Yup
<3> write it in an e-mail and send it
<5> then how do you portably write code that does byte-level operations?
<3> and find a way to merge algortihms
<1> Cowmoo it can... though I use the stuff in <limits> instead of those old defines these days
<5> I suppose better question is, what do you do when you need exactly 1 byte = 8 bits
<1> nush_ what compiler are you using?
<4> g++
<4> :/
<1> nush_ and you MIGHT want to put an extern on line 235
<4> okay let me look! thx
<7> Just ask! Easy : why Dev C++ isn't accepting "cin >> n1 >> n2 >> n3 >> n4..." ?
<1> oh, and do NOT put using namespace std; in a header file
<1> P-Brasil long time no type... and it does
<8> oh sweet, it's the "nah nah nah nah nah nah nah nah leader!" episode
<8> one of the best
<7> Yeah....i am programming better actually. But i'm making old programs to remember old concepts.
<1> so what makes you beleive that DevC++ doesn't accept that?
<4> vawjrwrk: same
<4> I am thinking about changing everything in node.h/cpp into a void * instead of hashtable *
<7> Because , last year i did something like that in the Borland C++ Builder , and it worked
<4> and then just casting it into a hashtable in the main.cpp file when i need to
<1> nush_ why the **** you want to mess with void*?
<1> you realize, don't you that ever "cast" is a lie to the compiler
<1> lemme guess, you learned C first
<4> vawjrwrk: because I don't know why or how to fix this problem. I tink its because main.cpp uses the hashtable and includes Nodes.h and Nodes also use the hashtable
<1> P-Brasil "isn't accepting" is a rather meaningless thing to tell us
<4> all I want is a Hashtable that has objects that store pointers to more hashtables. That is all I want!
<4> is it too much to ask for? :/
<1> possibly
<4> vawjrwrk: why?
<7> Okay. I will explain. This is a program from the 1st period. Actually i'm in the 2nd. I have to type nine numbers at same time...i know about multiple cin. What happens is that i type the numbers,and they don't appear at cout
<1> what HashTable are you using?
<4> the main hashtable stores node objects, and the node objects has a pointer that can point to another hashtable of node objects
<4> vawjrwrk: hashtable froma c++ book
<4> http://www.cs.fiu.edu/~weiss/dsaa_c++3/code/QuadraticProbing.h
<1> so what does IT say about having a hastable of pointers to hashtables?
<4> it says nothing! it was just for ints/strings and I added a hash function to work with nodes
<5> lol
<1> I don't see anywhere that your Node has a pointer to a hashtable
<9> how can i define a template method, which actually returns that kind of variable? something like inline T clamp<cl*** T>(T in, T min,T max)
<9> (that doesn't work)
<4> well that's cuz i changed it to void *
<4> but before it was Hashtable<node>
<5> template<typename T> T clamp(T var1, T var2);
<9> 10
<9> x
<1> nush_ and if you're having problems w/ void* then perhaps you should rethink using thim
<7> vawjrwrk: and i am not using old headers anymore :)
<4> well I changed to node * in an attempt to get away from multiple declerations
<4> vawjrwrk: what do you suggest i do?
<1> MrMuscolo template<typenameT> inline T clamp(T const& in, T const& min, T const& max)
<1> MrMuscolo looks like you need a good book on C++
<1> and p***ing by value can get rather expensive
<1> P-Brasil specifically what is DevC++ bitching about??
<7> vawjrwrk: i solved
<5> unless it's primitives only
<7> it was just something i forgot
<7> that i was looking in a e-book here,and found out
<7> cin will read until it finds a whitespace
<1> it ignores whitespace
<1> to start (unless told otherwise)
<4> vawjrwrk: I also have a version that uses STL hash_map, but i get another strange error in that one
<0> stringstream should have had an easier way to indicate what to use as a separator
<1> there is no STL hash_map
<1> nush_ what's the actual problem you're trying to solve
<0> vawjrwrk, whats the most efficient way to reset all values of a vector to 0
<0> its a big vector
<1> fill
<0> ok
<1> or perhaps fill_n
<0> yes yes
<1> not sure which would be better
<0> basically
<0> I have a 10000 value vector of doubles
<0> sorry thats 100000
<0> and only some indices do I set to a value
Return to
#c++
or
Go to some related
logs:
RELATIONS.MP3 ERIKA
#mirc
#AllNiteCafe
shaz rookiller
#c++
#linux
#php
t
#MissKitten
il-pitiross
|
|