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