Jump to content

skerry2006aj

  • Posts

    33
  • Joined

  • Last visited

  • Country

    Romania

Retained

  • Member Title
    - Software Developer -

Personal Information

  • Location
    Bucharest, Romania
  1. I've been working on something really interesting lately! By the way, check our page for our Visual Novel projects! https://www.facebook.com/seventh.heart.studios/ With the data we have so far, we should be able to construct an algorithm that should reduce clipping. In theory, we should be able to do this by constructing a soft-relaxed fast reacting filter that will detect edge tension (clipping) on the audio plot then erasing / re-modelling the curve shape / peak using a custom made algorithm. As many of us know for sure, it's not possible to eliminate clipping once it occurred. This is because we're talking about information that literally isn't there/ can't be recovered because of not existing there at all. Now the main idea we're working on is not going to solve that problem, neither are we going to say it's possible. But it is possible to alleviate the problem. I've managed to alleviate clipping from a voice sample by editing the file in audacity, shaped over the idea of the algorithm. The main problem is how would such an algorithm react in real life conditions and when it's processing real music. If it's too strict - the music will become soft, it will sound blurry and have a ton of noise / strange artifacts. If it's too soft - it won't be very effective and won't solve too many problems. The problem does not consist in building the algorithm, the problem is in constructing the clipping recognition algorithm - for this I'll need hundreds of tests and samples. But it's one heck of a hobby and I love playing with this. There is also the problem posed by processing everything in real time - or at least without taking too much time. While we can build a farm of computers / use a few XEON CPUs + 4X GTX 1080 to do this, the point is in bringing those features to the average user, even smartphone users - so we construct the algorithms to work for most people. It's possible I'll decide to make a super enhanced version to be used for actually remastering music in studios, which will need a ton more processing, for a little bump in quality, but I started with the idea of bringing a free ideal music playback software to users. In the meanwhile, as I stated earlier, I'm working on a few commercial projects, you can check out the pages for those projects. The software for music playback will be free - this is the main reason I'm making it. But I need to support development, and I'll be using your help of those projects to advance with my music playback software (development for it costs a ton of time and resources!). Check those links, like, subscribe, and most important of all, have fun! https://www.facebook.com/seventh.heart.studios/ https://twitter.com/ChronoStudios https://twitter.com/7heartstudios
  2. Yes! I have been working a lot on the project, and hit a wall where I need to invest money in it. I lately need much more hardware to be able to study in detail the effects of the algorithms on music, and I also need to assemble a small team as I get overwhelmed by how much code there is to write. Optimizing the software to run very very fast, have very good results, and to have the possibility to run on a mobile hardware is becoming harder than I thought, but I will not be lazy, I have to make it work. In programming, there is no impossible, there is just the principle of giving enough work hours and having enough knowledge to work. This being said, my actual machine, fast as it is, takes me 1 to 5 minutes to analyse short samples at the resolution I need them analysed. The processing part works as it should, but I have to always track how the algorithms interact with every sound, and a musical piece. The idea is that it is extremely easy to code it to do magic for drums, but then it makes the voices sound unnatural. To avoid this, I make the algorithm be an artificial intelligence and learn and adapt to what the sound is. The algorithm is able to recognize what a sound means in real world, regardless of how it was recorded, based on very well defined learning algorithms, and thus the final software is able to make all music sound amazing, and not just one or two instruments, or music recorded in just one or two ways. It is a lot of work to be done, as artificial intelligence is much more complex than it seems, and it needs hundreds of hours of development to work as it is supposed to work. Right now I am working to be able to invest in the project. I am programming, writing and directing AAA class Visual Novels. I can work pretty fast, and working in python programming environment is easy compared to what C# + C++ was doing to me. After getting into this, I learned how to program artificial intelligence in python. Right now I am considering restructuring some of the coding in python, as I am getting used to it much faster than I thought, and writing the critical parts in C++, but I guess that knowing how perfectionist I am when it comes to this project, I might end up using C++ alone to have the best level of control. Either way, it is going great, but most developments and very amazing discoveries are laboratory level at this moment, but I will be sure to show interesting results as soon as they are out of laboratory level. Working with huge resolution and precision data needs a lot of processing power to interpret, then to transform into something that can be played and then making acoustic testing on how the changes affected a real life performance. I also need to constantly buy more recording / playback equipment and sonic isolation for the rooms the testings are being conducted in. I wanted to just post about the Visual Novels projects when the projects were much closer to being ready, at least the demo being ready. There are a few very juicy secrets, and some of them will be leaked, but to give a little leak in here, exclusively first on computer audiophile: I plan for all soundtracks to be in ogg -q10, and the image part will also have something very special for all fidelity enthusiasts out there. While for just the audio playback project, it would probably be simpler to just crowdfund it or sell it to a bigger company who would just program it under my guidance, I will have a pre order option, but when I have a fully working demo for all platforms ready. I want people to be content with the software, to preorder. Even if it takes a year by that time, I want to be able to give real tangible results to people following the project, and taking an interest in it. And about a bigger company buying it, I am not entirely opposed to the idea, but I am not entirely in love with the idea, I would like to be able to control the final product, and what it comes out, and what it is after the highest standards, and make no sacrifices, or introduce any gimmicks. If any company would comply to this, they could try to, but the last thing I want to make is an average product, I want this product to be "The ideal music playback software", not just another music playback software. In the meantime, I am also working on wireless data compression algorithms, that would be able to losslessly send data between two points. This could be extremely useful in having real life quality phone calls, or having true lossless music online, on an internet enabled player, but without having to download 40Mb for a single song. This compression algorithm is meant to keep information lossess, and to apply all algorithms created by me, on a mobile player. So far, I think that high end smartphones will be able to pull this off, while mid ranged smartphones and players will also be able to do this, but not right now, maybe one or two generations later, when the microprocessors inside and GPUs permit the algorithms to work at full capacity, without running very hot, or consuming a lot of battery.
  3. I am not very accustomed to Last.fm protocols, but I am thinking that adding support for it is possible, and I will add it. Wave form, spectrograms, dynamic range analyzer, some skins, VU meters, and other visual effects are on my list. I spend time studying spectrograms created by a few problems, while listening to certain songs, and find studying them a fun hobby. Lyrics support is a must for a good music player. In fact, if time permits, I would like to add support for an automated lyrics finder, and use machine learning to build an option within the software to automatically search databases of lyrics and display the lyrics of the song that is played. A general method for this should cover 99.9% of the songs out there, and all songs that exist within a clear database. The visual experience, UI, and general aspect and usage of the program is one of the first priorities for me, as I want users to use the program because they actually like it from all points of view, not just because it sounds good, or does something good. I aim to build it to do everything in music experience good.
  4. UPNP is an important thing for many, and there is enough demand for it as a feature, I will make sure to have good implementation of it!
  5. Yeah, when playing a file within a folder, all files that follow it in alphabetical order will be played next. After a folder is finished, the next folder alphabetically will start playing and so on. You can change the song at any time, and I plan to integrate usage of keyboard multimedia keys. This seems like it has been ignored to be properly supported lately, but I think that some people might find it helpful.
  6. I will look into it, and I will try to think of something to please users accustomed to this method of sorting playlists.
  7. Exactly this is the way I intent my software to be. I want it to support it all transparently, to please every category of users. I want to create a software that will be as universal as possible.
  8. All tags that are supported by MP3Tag will be supported. I will have a Title (song name), Album artist, Composer, Disc number, Comment, Date, Track number, Genre, Album, Artist. Adding more tags than are supported by other software could create problems, like portable media players not being able to play your files. I will look into how metadata is recorded and used. If I will be able to (not create compatibility problems with your files) I will add Recording label, Orchestra, Place ( Recording place). But I have to study this further. I think that iTunes have some sort of patent on that particular way to use columns, but I will create something that will help people who browse by tags. I have a few ideas on paper, I will test and implement the one who lets you browse with most ease, and is the most intuitive.
  9. Library interface will allow folder path view in column. If you mean having main directory, then expand every sub directory, this is simple to add to my software, and is not a problem, you will be able to double click from that point and play a song. I have not thought that this would be more wanted than having a column with file path. I do not have that many tags, and I like to define a very exact folder tree. I have been using Fiio X5 and X5II as portable players, and they work well with my tidy folder structure. For my needs, I have tags defined only for Title, Artist, Album, Year, Genre, and Track. I personally prefer to have .jpg high quality cover art within folder, instead of having it attached as metadata. On the other hand, I understand very well that other people have other preferences, so I will try to make my software be best fit for every possible user. Having a simple method to customize it, but making customization reliable should do the trick. I acknowledge UPnP as being important, and will offer support for it. I think that people who like sound quality will be looking forward to the transient recovery algorithm I am working on. It is based upon machine learning, and will provide one of the best results in sound to date. I also plan to make a Linux version.
  10. Thanks for your answer! I have thought about this, and I actually plan to do a feature that could improve the situation for most people who want to have a dedicated software for music playback. I plan to add a pure music mode to the software. Like this, it would satisfy both audiophiles and music enthusiasts and videophiles. After I have played around with Fidelizer for windows, I think that adding a pure music mode would improve audio quality. I will need GPU processing for my Transient Recovery Algorithm, and having video play at the same time will be possible but will depend on what madVR settings people have for video playback, and on how resource hungry the final release of my algorithm will be. Until now, I can say that it is hard to process, as are most machine learning algorithms. madVR is very nice for video playback and I want to offer easier ways for people to use it, this is why I plan to include options for it. (It took me a while before understanding it and being able to make a configuration that worked for me) I am still considering this, and it would be pretty interesting for some, but would take a bit more coding to do, to actually use a SLI GPU configuration (for people who have one) and process video playback in one GPU, and my algorithm in the other GPU. I cannot make an assertion of how many people have SLI setups, are audio enthusiasts and also want the best audio quality possible, and best video quality possible.
  11. I posted a few new updates to my project, and a small guide for how to use mpc-hc with madVR. It is the best experience for video watching that I found, and it is free. It inspired me to want to add madVR support for my software, for video playback. It seems to me as if most people who are into music might want to have in the same package the best experience with video watching too. I invite you to read more on my project's page The ideal music playback software. Great sound, intuitive and elegant user interface.
  12. I have been considering lately that people might want to use the software for video watching too. I, myself, like to watch some AMV (anime music videos) from time to time, so I think that I am going to try to encode something for video too, and I want to also imlpement one of those motion interpolation algorithms for video too. Keeping in mind that my project is designed to be free, I might be interested in working with SVP team at a point, because I am a fan of their works, and I am using their software quite often.
  13. Hello! Thank you for asking! I sure do plan to include support for multichannel music. I have not said anything about it because at the moment I do not own any multichannel equipment. So, just at the moment I cannot test how multichannel will sound like after implementations. But I think that I can optimize something for multichannel to make it sound great! After getting more financial resources for my project and getting better equipment, this is something that I will be sure to add to my project.
  14. After this has been requested, I had decided to add support for .WV files (Wavpack). As always, please request any features that you want, and I will try to implement them ! Gheorghe Dobrescu
×
×
  • Create New...