TAK 1.0 - Final release of the new lossless codec, offers superior performance Options

[TBeck]

TAK 1.0 - Final release

It's been a long time...

About

TAK is a lossless audio compressor, similar to FLAC, WavPack and Monkey's Audio. On average, lossless compression reduces the file size to about 50 percent of the original size; however the compression can vary between approximately 30 and 70 percent, depending on the file. Decompression restores a bit identical copy of the original audio data (that's why we call it lossless).

My goal was to develop a compressor which combines good compression with optimal decoding speeds. On average, the current implementation should match the compression efficiency of Monkey's Audio High, while achieving decompression speeds similar to FLAC.

Features

- Good compression. The strongest mode (Extra) is on par with Monkey's Audio High and OptimFrog Normal. TAK's fastest mode (Turbo) easily outperforms FLAC's strongest mode. This classification is based upon the evaluation of hundreds of files of different genres; it may not be the case for every file that you compress.
- Fast compression speed. Currently I know of no other compressor which compresses faster than TAK's Turbo and Fast modes, while providing similar compression ratios.
- Very fast decompression speed. The decoding speeds achieved are similar to, and often surpassing, those achieved by FLAC.
- Support for any common audio format (not fully implemented).
- Streaming support. An info frame, which contains all the information required to decompress a file, is inserted into the compressed audio data every 2 seconds.
- Error tolerance. A single bit error will never affect more than 250 ms of the audio data, as the compressed data is being stored in independent frames no more than that duration. The decoder is able to decompress even badly damaged files, by removing corrupt frames or replacing them with silence.
- Error detection. Each single frame is protected by a 24-bit checksum (CRC).
- Fast and sample accurate seeking. The file header contains a seektable with seekpoints (at 1 second intervals). In the absence of the seektable seeking can still be achieved by using the sync codes of the frame headers and the (optional) relative offset values contained in each info frame, which are pointing to the previous and next info frame.
- Meta data. A flexible and extensible structure in which to embed non audio data.

Changes

Improved:

- Synthetic Soul was so kind to correct my bad english in the ReadMe.

Fixed:

- Decoding with the command line version: if you specified wildcards for the file selection and the source directory contained only 1 TAK file, the decoder always threw the message 'File already exists'. Even worse: using the overwrite option in this situation led to deletion of the compressed source file!

- Usually the decoder ignores any data appended to the file end of the compressed file (for instance APEv2 tags). But it failed, if the file size was an integer multiple of the frame size (in samples).

Both bugs affected only the decoder, therefore there is no need to re-encode TAK files created with beta 2. Files created with earlier versions than beta 2 should be replaced.

Thank you

Thank you to all the members who have supported the development of TAK!

Firstly, you have provided encouragement to build yet another lossless audio compressor. Otherwise I would have dropped the whole project!

You helped to evaluate and optimize TAK. There have been 13 releases over a period of 8 months before the alpha version, and all of them have been evaluated by you!

Special thanks to the following members (in alphabetical order):

Destroid
Josef Pohm
Synthetic Soul

Download

The final can be downloaded from rarewares.org (Thanks to rjamorim!):

TAK 1.0 (Final)

Future

Some important features are missing in this release. Please find below my current to-do list:

- Plugins for other applications, such as Winamp or foobar2000 (a Winamp plugin is already close to completion).
- An SDK for other developers.
- Internal tagging.
- Unicode support.
- Piping support.
- MD5 audio checksums for verification and identification.
- A german version.
- Embedded cue sheets.
- Embedded cover art.
- Multi channel audio.
- Even more speed and better compression.

This post has been edited by TBeck: Jan 30 2007, 05:18

[Mangix]

the wav file that i made came from an mp3(wanted to test out TAK's prefilter). but for some reason, TAK doesn't like it.

what's also weird is that if i use LAME to decode the mp3, the wav file will play nicely with TAK.