Sample rate and bit rate Part 2
Listen and compare
Why don’t you really ask? In my memory, when I checked it in the past, I remember that it was difficult to distinguish it from the original sound (PCM) at 128 kbps of AAC under the conditions in the table above. I think this varies from person to person, and although I am involved with the audio and sound, I am aware that my ears are not a big problem, so even at a slightly higher rate, it is the same as the sound. original. I’m sure there are people who can tell the difference. At the low 32 kbps, you can clearly see the difference in sound quality. In terms of music, you can understand the metallic sound of the drum hi-hat.
Personally, I think that 44.1 Hz 16-bit (stereo) music CDs can be saved even at 128 kbps (1/10 compression or less) without losing sound quality. About 128 kbps is enough for my ears for both MP3 and AAC.
The bit rate is the compression rate
What happens if you set the encoding bit rate to 256 kbps for 16 kHz audio (monaural with 16 quantization bits)? .. .. Since the compression rate is 100%, it will be the same as the original sound. The sound quality should be the same as the original sound, but it may cause strange behavior depending on the encoders that are available for free (a configuration error may occur).
Sampling frequency Number of quantization bits Number of channels Original sound bit rate (PCM) Remarks
32 kHz 16 1 512 kbps Super Wide Band
24 kHz 16 1 384 kbps
16 kHz 16 1 256 kbps Broadband
8 kHz 16 1 128 kbps Narrowband
Regarding lossy compression of AAC and MP3, I think it is the result of research on how to encode at a low rate, so I personally think that setting a bitrate of 50% or more is not good. Lossless is recommended for compression ratios around 50% (lossless compression, MPEG-4 ALS, etc.). If you only think about saving, even if you compress it as is in PCM, it seems like it’s about half for audio with quiet sections. For lossy compression AAC, MP3, etc., if sound quality is important, about 15-20%, and if high compression is important, about 10% is sufficient sound quality.
Also, for audio purposes less than 10% and 5% is fine, but for audio it is recommended to lower the sample rate rather than suppress the bit rate to 48 kHz or 44.1 kHz (8 kHz or 16 kHz).
Stereo M / S (middle side)
The left and right signals are sum / difference signals. When encoding the sum signal (L + R) and the difference signal (LR) of both channels, the code is used when the correlation between channels is high, such as in stereo. The conversion efficiency is improved. For example, you can improve the coding efficiency of musical voices (L / R in phase, same amplitude).
Intensity stereo
When listening to high frequencies, the bit rate is reduced by combining the high frequency information (quantization coefficient) into one using the property that it is more susceptible to loudness than the L / R time difference.
In the end
Although bit rate may seem like a measure of sound quality, the digital audio field does not specify an encoded bit rate that exceeds the original sound bit rate. In short, I think it is important to use the proper bitrate for each encoder (encoder).
