Fork: 0
kalle / monitord
Transfer to URL with SHA Find file
Newer
Older
monitord / lame-3.97 / TODO
@root root on 23 Jan 2012 4 KB Migration from SVN revision 455
Raw Blame History 
  1. 1. bug in resample code:  downsampling from 44101 to 44100 causes
  2.    a seg fault.  Workaround in place for now:  resampling disabled 
  3.    if input/output samplerates agree to 4 digits.  
  4.  
  5.  
  6.  
  7. 2.  high bitrate encodings have trouble on some hardware players.
  8. Track this down.  Probably caused by --strictly-enforce-ISO and
  9. IXMAX_VAL.  Try setting IXMAX_VAL back to 8191 and/or
  10. maxmp3buf=8*960 to see if there is a working combination.
  11.  
  12. note: one of the decoder bugs was identified. It is caused by using
  13. different block sizes on both channels. A parameter need to be
  14. added to Lame to handle workarounds.
  15.  
  16.  
  17. 3 frontend: code is a complete mess. But it has so many debugged
  18.   features it will be a lot of work to re-write.
  19.  
  20.  
  21. 4. MSVC project files.  It would be nice to create a working
  22.    MSVC6 workspace, which included all the projects as possible
  23.    targets:
  24.           lame.exe
  25.           mp3x.exe   (require GTK libs)
  26.           lame_enc.dll   
  27.           ACM codec
  28.           directshow codec
  29.  
  30.    I think the only MSVC5 project that we need to preserve is
  31.    for lame_enc.dll, since Albert Faber (still?) doesn't use VC6?
  32.    But no reason we cant have VC5 and VC6 project files for the dll.
  33.  
  34.    
  35.  
  36.    
  37.  
  38.  
  39. 5.
  40. NOGAP encoding:
  41.  
  42. -nogap:  more testing, fix options, test id3 tags?
  43. Can we change id3 tags without reseting the encoder??
  44. At the end of encoding 1.wav, call lame_get_mf_samples_to_encode()
  45. to find the number of non encoded buffered PCM samples.  Then
  46. encode samples from 2.wav until these PCM samples have been
  47. encoded, *THEN* call lame_encode_flush_nogap() and close
  48. out file 1.mp3.
  49.  
  50.  
  51. NOGAP decoding:  
  52. lame --decode --nogap file1.mp3 file2.mp3 file3.mp3
  53. should also work.  What needs to be done:
  54. get_audio.c:  We need a way to open a second mp3 file, without
  55.               calling lame_decode_init() and reinitializing mpglib.
  56.               And the mpglib needs to know to look for new Xing
  57.               tags at the beginning of file2.mp3 and file3.mp3.
  58.  
  59.  
  60.  
  61. 6.
  62. Does stdin work when LAME is compiled to use libsndfile? 
  63. (new version of libsndfile will support this - try this out)
  64.  
  65.  
  66. 7. 
  67. LAME has problems with pure DC input.  i.e. a square wave with
  68. a frequency well below 20 Hz.  Not very important, but it should
  69. be fixed.
  70.  
  71.  
  72. 8.
  73. mgplib has bugs with i-stereo.  flag denoting invalid
  74. i-stereo value (= frame is m/s stereo) is not correct.  
  75.  
  76. 9.
  77. lowpass filter: for M/S stereo, use more filtering for the side
  78. channel, less filtering for mid channel.  We need to first replace
  79. the polyphase filter with an FIR lowpass filter with finer frequency
  80. resolution before implementing this. 
  81.  
  82. 10. 
  83. LAME has a 31 point FIR filter used for resampling, which
  84. can also be used as a lowpass.  When resampling is done, 
  85. use that filter to also lowpass instead of the polyphase filter.
  86.  
  87. 11.
  88. Even when resampling is not needed, should we use an FIR filter
  89. for the lowpass?  If it is not too much slower, yes.  If it
  90. is slower, then it should be an option since it will produce
  91. higher quality.  
  92.  
  93. 12.
  94. We should consider moving the experts options from the *long
  95. help* text into an *experts only* help text. The average Joe gets
  96. knocked down by the huge number of possibilities to setup lame.
  97.  
  98.  
  99.  
  100.  
  101. 50.
  102. Better tonality estimation.
  103. Gpsycho uses predictability, and so needs a delay to detect the tonality
  104. of a sound.
  105. Nspsytune seems to miss tonals when several of them are too narrow.
  106. We would probably need the best of both.
  107.  
  108.  
  109.  
  110.  
  111. 60.
  112. Different ATH handling for sfb21. We are using the minimum value of ath
  113. in each whole sfb. in sfb21 this leads to very high bitrates.
  114. We could perhaps use 2 or 3 ath partitions in sfb21
  115.  
  116. note: partially done
  117.  
  118.  
  119.  
  120. 70.
  121. Use mixed blocks.
  122.  
  123.  
  124.  
  125.  
  126. 90.
  127. Use intensity stereo. This is a must-have for low bitrates, but if the
  128. algorythm is very good it could also be used in every case.
  129. Note: mpg123 (and all derivatives, like xmms and lame/mpglib)
  130. have bugs in the intensity stereo decoding.  Bugs have been there
  131. for years since there are very few intensity stereo mp3's out there.
  132.  
  133.  
  134.  
  135.  
  136.  
  137. 998.
  138. Merge GOGO's fast assembler routines.
  139.  
  140.  
  141.  
  142.  
  143. 999.
  144. It would be nice to save some information whilst encoding
  145. a: wave -> mp3
  146. 	a RIFF/wave can contain LIST chunks with information
  147. 	about author, title, etc. 
  148. 	==> could go into TAG fields of resulting mp3
  149. b: mp3 -> mp3
  150. 	==> we could copy the TAG directly
  151. c: mp3 -> wave
  152. 	==> copy TAG into LIST chunk
  153.  
  154.  
  155.  
  156. 1500.
  157. Integrate plusV extensions
  158.  
  159.  
  160.  
  161. 2000.
  162. To be able to encode as fast as FastEnc