This page explains the practical difference between ordinary LAME VBR, minimum VBR, strict cVBR and the cVBRb bitrate boost levels. The results are based on one lossless WAV test input encoded with the same cVBRb-capable LAME 3.101 beta 3 build.
Ordinary minimum VBR is not a strict frame-by-frame floor. In this test, -V0 -b 192 still produced 17 frames at 32 kbps, most likely during silence or very low-complexity passages.
Strict cVBR removes those below-minimum frames. With --vbr-min-strict, every logged audio frame stayed at or above the requested 192 kbps minimum.
cVBRb adds optional extra margin. The boost levels progressively shift more frames toward 256 and 320 kbps.
The original motivation for cVBR was listening-based. During testing, some VBR encodes appeared to produce slight artifacts or distortion when bitrate changes were extreme. That kind of difference can be difficult to prove objectively, and many listeners may not hear a clear difference at high-quality VBR settings.
What can be shown clearly is the encoder behaviour. Ordinary minimum VBR does not necessarily enforce the selected minimum on every frame. Strict cVBR changes that behaviour by applying a stricter minimum bitrate floor across the stream.
cVBR is therefore best understood as an advanced control option: not a magic quality switch, but a more conservative and predictable form of VBR for users who prefer avoiding very low-bitrate frames and extreme bitrate swings.
-b 192. This requests a minimum bitrate, but the test shows it is not an absolute floor in all cases.--vbr-min-strict, enforcing the requested minimum bitrate more strictly.All modes used -V0. Minimum and cVBR modes used -b 192. The source audio is not distributed; users can repeat the test with their own legally obtained WAV or FLAC files.
| Mode | Command options | Size | Avg kbps | 32 | 192 | 224 | 256 | 320 | Time |
|---|---|---|---|---|---|---|---|---|---|
| VBR | -V0 |
103.82 MiB | 263.1 | 17 | 2,543 | 30,705 | 61,531 | 32,019 | 26s |
| Minimum VBR | -V0 -b 192 |
103.82 MiB | 263.1 | 17 | 2,599 | 30,678 | 61,530 | 32,021 | 26s |
| Strict cVBR | -V0 -b 192 --vbr-min-strict |
103.83 MiB | 263.1 | 0 | 2,616 | 30,678 | 61,530 | 32,021 | 27s |
| cVBRb light | -V0 -b 192 --vbr-min-strict --bitrate-boost=1 |
105.77 MiB | 268.0 | 0 | 2,044 | 22,823 | 64,673 | 37,305 | 26s |
| cVBRb medium | -V0 -b 192 --vbr-min-strict --bitrate-boost=2 |
107.94 MiB | 273.5 | 0 | 1,641 | 15,278 | 65,901 | 44,025 | 26s |
| cVBRb aggressive | -V0 -b 192 --vbr-min-strict --bitrate-boost=3 |
110.56 MiB | 280.2 | 0 | 1,291 | 8,401 | 63,747 | 53,406 | 27s |
-b 192 alone is not strictNormal -V0 -b 192 still logged 17 frames at 32 kbps. Those frames are probably silence or near-silence, but they show that ordinary minimum VBR behaves as a minimum target rather than an absolute frame-by-frame floor.
Adding --vbr-min-strict removed the below-minimum 32 kbps frames. On this sample, the lowest logged bitrate became 192 kbps.
The boost modes progressively reduced 224 kbps frames and increased 256/320 kbps frames. Average bitrate rose from 263.1 kbps to 268.0, 273.5 and 280.2 kbps.
The tested encodes completed in roughly 26–27 seconds, so the stricter modes did not introduce a large speed penalty in this run.
At high-quality VBR settings, many listeners may not hear a clear difference between ordinary VBR, cVBR and cVBRb. That is expected. MP3 encoding is psychoacoustic, and a well-tuned VBR encode can already be transparent on much material.
The comparison should therefore be read as evidence of bitrate-control behaviour, not as proof that every listener will hear an improvement. cVBR and cVBRb are offered for users who value stricter control, extra margin, or want to experiment with a more conservative encode.
Use any lossless WAV or FLAC source that you legally own. The source audio used for this page is not distributed.
lame_cVBRb_x64.exe -V0 input.wav output-vbr.mp3
lame_cVBRb_x64.exe -V0 -b 192 input.wav output-min-vbr.mp3
lame_cVBRb_x64.exe -V0 -b 192 --vbr-min-strict input.wav output-cvbr.mp3
lame_cVBRb_x64.exe -V0 -b 192 --vbr-min-strict --bitrate-boost=1 input.wav output-cvbrb1.mp3
lame_cVBRb_x64.exe -V0 -b 192 --vbr-min-strict --bitrate-boost=2 input.wav output-cvbrb2.mp3
lame_cVBRb_x64.exe -V0 -b 192 --vbr-min-strict --bitrate-boost=3 input.wav output-cvbrb3.mp3For listening quality, use blind ABX testing with your own material. Frame distribution and bitrate tables can show how the encoder behaves, but they do not guarantee what every listener will hear.