Recherche avancée

Sur d’autres sites (4292)

• Chrome’s New Audio Notifier

  30 janvier 2014, par Multimedia Mike — General

  Version 32 of Google’s Chrome web browser introduced this nifty feature :

  
  
  

  When a browser tab has an element that is producing audio, the browser’s tab shows the above audio notification icon to inform the user. I have seen that people have a few questions about this, specifically :

  1. How does this feature work ?
  2. Why wasn’t this done sooner ?
  3. Are other browsers going to follow suit ?

  Short answers : 1) Chrome offers a new plugin API that the Flash Player is now using, as are Chrome’s internal media playing facilities ; 2) this feature was contingent on the new plugin infrastructure mentioned in the previous answer ; 3) other browsers would require the same infrastructure support.

  Longer answers follow…

  Plugin History
  Plugins were originally based on the Netscape Plugin API. This was developed in the early 1990s in order to support embedding PDFs into the Netscape web browser. The NPAPI does things like providing graphics contexts for drawing and input processing, and mediate network requests through the browser’s network facilities.

  What NPAPI doesn’t do is handle audio. In the early-mid 1990s, audio support was not a widespread consideration in the consumer PC arena. Due to the lack of audio API support, if a plugin wanted to play audio, it had to go outside of the plugin framework.

  
  
  

  There are a few downsides to this approach :

  • If a plugin wants to play audio, it needs to access unique audio APIs on each supported platform. One of the most famous things I’ve ever written deals concerns this nightmare on Linux. (The picture worth a thousand words.)
  • Plugin necessarily needs free unrestricted access to system facilities, i.e., security measures like sandboxing become more difficult without restricting functionality.
  • Since the browser doesn’t mediate access to the audio APIs, the browser can’t reasonably be expected to know when a plugin is accessing the audio resources.

  So that last item hopefully answers the question of why it has been so difficult for NPAPI-supporting browsers to implement what seems like it would be simple functionality, like implementing a per-tab audio notifier.

  Plugin Future
  Since Google released Chrome in an effort to facilitate advancements on the client side of the internet, they have made numerous efforts to modernize various legacy aspects of web technology. These efforts include the SPDY protocol, Native Client, WebM/WebP, and something call the Pepper Plugin API (PPAPI). This is a more modern take on the classic plugin architecture to supplant the aging NPAPI :

  
  
  

  Right away, we see that the job of the plugin writer is greatly simplified. Where was this API years ago when I was writing my API jungle piece ?

  The Linux version of Chrome was apparently the first version that packaged the Pepper version of the Flash Player (doing so fixed an obnoxious bug in the Linux Flash Player interaction with GTK). Now, it looks like Windows and Mac have followed suit. Digging into the Chrome directory on a Windows 7 installation :

  AppData\Local\Google\Chrome\Application[version]\PepperFlash\pepflashplayer.dll

  This directory exists for version 31 as well, which is still hanging around my system.

  So, to re-iterate : Chrome has a new plugin API that plugins use to access the audio API. Chrome knows when the API is accessed and that allows the browser to display the audio notifier on a tab.

  Other Browsers
  What about other browsers ? “Mozilla is not interested in or working on Pepper at this time. See the Chrome Pepper pages.”

• Sequencing MIDI From A Chiptune

  28 avril 2013, par Multimedia Mike — Outlandish Brainstorms

  The feature requests for my game music appreciation website project continue to pour in. Many of them take the form of “please add player support for system XYZ and the chiptune library to go with it.” Most of these requests are A) plausible, and B) in process. I have also received recommendations for UI improvements which I take under consideration. Then there are the numerous requests to port everything from Native Client to JavaScript so that it will work everywhere, even on mobile, a notion which might take a separate post to debunk entirely.

  But here’s an interesting request about which I would like to speculate : Automatically convert a chiptune into a MIDI file. I immediately wanted to dismiss it as impossible or highly implausible. But, as is my habit, I started pondering the concept a little more carefully and decided that there’s an outside chance of getting some part of the idea to work.

  Intro to MIDI
  MIDI stands for Musical Instrument Digital Interface. It’s a standard musical interchange format and allows music instruments and computers to exchange musical information. The file interchange format bears the extension .mid and contains a sequence of numbers that translate into commands separated by time deltas. E.g. : turn key on (this note, this velocity) ; wait x ticks ; turn key off ; wait y ticks ; etc. I’m vastly oversimplifying, as usual.

  MIDI fascinated me back in the days of dialup internet and discrete sound cards (see also my write-up on the Gravis Ultrasound). Typical song-length MIDI files often ranged from a few kilobytes to a few 10s of kilobytes. They were significantly smaller than the MOD et al. family of tracker music formats mostly by virtue of the fact that MIDI files aren’t burdened by transporting digital audio samples.
  
  I know I’m missing a lot of details. I haven’t dealt much with MIDI in the past… 15 years or so (ever since computer audio became a blur of MP3 and AAC audio). But I’m led to believe it’s still relevant. The individual who requested this feature expressed an interest in being able to import the sequenced data into any of the many music programs that can interpret .mid files.

  The Pitch
  To limit the scope, let’s focus on music that comes from the 8-bit Nintendo Entertainment System or the original Game Boy. The former features 2 square wave channels, a triangle wave, a noise channel, and a limited digital channel. The latter creates music via 2 square waves, a wave channel, and a noise channel. The roles that these various channels usually play typically break down as : square waves represent the primary melody, triangle wave is used to simulate a bass line, noise channel approximates a variety of percussive sounds, and the DPCM/wave channels are fairly free-form. They can have random game sound effects or, if they are to assist in the music, are often used for more authentic percussive sounds.

  The various channels are controlled via an assortment of memory-mapped hardware registers. These registers are fed values such as frequency, volume, and duty cycle. My idea is to modify the music playback engine to track when various events occur. Whenever a channel is turned on or off, that corresponds to a MIDI key on or off event. If a channel is already playing but a new frequency is written, that would likely count as a note change, so log a key off event followed by a new key on event.

  There is the major obstacle of what specific note is represented by a channel in a particular state. The MIDI standard defines 128 different notes spanning 11 octaves. Empirically, I wonder if I could create a table which maps the assorted frequencies to different MIDI notes ?

  I think this strategy would only work with the square and triangle waves. Noise and digital channels ? I’m not prepared to tackle that challenge.

  Prior Work ?
  I have to wonder if there is any existing work in this area. I’m certain that people have wanted to do this before ; I wonder if anyone has succeeded ?

  Just like reverse engineering a binary program entails trying to obtain a higher level abstraction of a program from a very low level representation, this challenge feels like reverse engineering a piece of music as it is being performed and automatically expressing it in a higher level form.

• The problems with wavelets

  27 février 2010, par Dark Shikari — DCT, Dirac, Snow, psychovisual optimizations, wavelets

  I have periodically noted in this blog and elsewhere various problems with wavelet compression, but many readers have requested that I write a more detailed post about it, so here it is.

  Wavelets have been researched for quite some time as a replacement for the standard discrete cosine transform used in most modern video compression. Their methodology is basically opposite : each coefficient in a DCT represents a constant pattern applied to the whole block, while each coefficient in a wavelet transform represents a single, localized pattern applied to a section of the block. Accordingly, wavelet transforms are usually very large with the intention of taking advantage of large-scale redundancy in an image. DCTs are usually quite small and are intended to cover areas of roughly uniform patterns and complexity.

  Both are complete transforms, offering equally accurate frequency-domain representations of pixel data. I won’t go into the mathematical details of each here ; the real question is whether one offers better compression opportunities for real-world video.

  DCT transforms, though it isn’t mathematically required, are usually found as block transforms, handling a single sharp-edged block of data. Accordingly, they usually need a deblocking filter to smooth the edges between DCT blocks. Wavelet transforms typically overlap, avoiding such a need. But because wavelets don’t cover a sharp-edged block of data, they don’t compress well when the predicted data is in the form of blocks.

  Thus motion compensation is usually performed as overlapped-block motion compensation (OBMC), in which every pixel is calculated by performing the motion compensation of a number of blocks and averaging the result based on the distance of those blocks from the current pixel. Another option, which can be combined with OBMC, is “mesh MC“, where every pixel gets its own motion vector, which is a weighted average of the closest nearby motion vectors. The end result of either is the elimination of sharp edges between blocks and better prediction, at the cost of greatly increased CPU requirements. For an overlap factor of 2, it’s 4 times the amount of motion compensation, plus the averaging step. With mesh MC, it’s even worse, with SIMD optimizations becoming nearly impossible.

  At this point, it would seem wavelets would have pretty big advantages : when used with OBMC, they have better inter prediction, eliminate the need for deblocking, and take advantage of larger-scale correlations. Why then hasn’t everyone switched over to wavelets then ? Dirac and Snow offer modern implementations. Yet despite decades of research, wavelets have consistently disappointed for image and video compression. It turns out there are a lot of serious practical issues with wavelets, many of which are open problems.

  1. No known method exists for efficient intra coding. H.264′s spatial intra prediction is extraordinarily powerful, but relies on knowing the exact decoded pixels to the top and left of the current block. Since there is no such boundary in overlapped-wavelet coding, such prediction is impossible. Newer intra prediction methods, such as markov-chain intra prediction, also seem to require an H.264-like situation with exactly-known neighboring pixels. Intra coding in wavelets is in the same state that DCT intra coding was in 20 years ago : the best known method was to simply transform the block with no prediction at all besides DC. NB : as described by Pengvado in the comments, the switching between inter and intra coding is potentially even more costly than the inefficient intra coding.

  2. Mixing partition sizes has serious practical problems. Because the overlap between two motion partitions depends on the partitions’ size, mixing block sizes becomes quite difficult to define. While in H.264 an smaller partition always gives equal or better compression than a larger one when one ignores the extra overhead, it is actually possible for a larger partition to win when using OBMC due to the larger overlap. All of this makes both the problem of defining the result of mixed block sizes and making decisions about them very difficult.

  Both Snow and Dirac offer variable block size, but the overlap amount is constant ; larger blocks serve only to save bits on motion vectors, not offer better overlap characteristics.

  3. Lack of spatial adaptive quantization. As shown in x264 with VAQ, and correspondingly in HCEnc’s implementation and Theora’s recent implementation, spatial adaptive quantization has staggeringly impressive (before, after) effects on visual quality. Only Dirac seems to have such a feature, and the encoder doesn’t even use it. No other wavelet formats (Snow, JPEG2K, etc) seem to have such a feature. This results in serious blurring problems in areas with subtle texture (as in the comparison below).

  4. Wavelets don’t seem to code visual energy effectively. Remember that a single coefficient in a DCT represents a pattern which applies across an entire block : this makes it very easy to create apparent “detail” with a DCT. Furthermore, the sharp edges of DCT blocks, despite being an apparent weakness, often result in a “fake sharpness” that can actually improve the visual appearance of videos, as was seen with Xvid. Thus wavelet codecs have a tendency to look much blurrier than DCT-based codecs, but since PSNR likes blur, this is often seen as a benefit during video compression research. Some of the consequences of these factors can be seen in this comparison ; somewhat outdated and not general-case, but which very effectively shows the difference in how wavelets handle sharp edges and subtle textures.

  Another problem that periodically crops up is the visual aliasing that tends to be associated with wavelets at lower bitrates. Standard wavelets effectively consist of a recursive function that upscales the coefficients coded by the previous level by a factor of 2 and then adds a new set of coefficients. If the upscaling algorithm is naive — as it often is, for the sake of speed — the result can look quite ugly, as if parts of the image were coded at a lower resolution and then badly scaled up. Of course, it looks like that because they were coded at a lower resolution and then badly scaled up.

  JPEG2000 is a classic example of wavelet failure : despite having more advanced entropy coding, being designed much later than JPEG, being much more computationally intensive, and having much better PSNR, comparisons have consistently shown it to be visually worse than JPEG at sane filesizes. Here’s an example from Wikipedia. By comparison, H.264′s intra coding, when used for still image compression, can beat JPEG by a factor of 2 or more (I’ll make a post on this later). With the various advancements in DCT intra coding since H.264, I suspect that a state-of-the-art DCT compressor could win by an even larger factor.

  Despite the promised benefits of wavelets, a wavelet encoder even close to competitive with x264 has yet to be created. With some tests even showing Dirac losing to Theora in visual comparisons, it’s clear that many problems remain to be solved before wavelets can eliminate the ugliness of block-based transforms once and for all.