the frequency response of vs1053/vs1063

[Churmenton]

hi all,
vs1053 encodes ogg audio format based documents,there are 6 level when samplerate is 44.1khz,if select the minimum level,the frequency response merely reach 13.3khz,if select the middle level,the frequency response merely reach 15khz,if select the bigest level,the frequency response merely reach 15.6khz.no method to make the frequency response reach or over 20khz.i wander if the frequency response of our vs1053/vs1063 chip can reach 20khz,if not,which chip can get that high frequency reponse.

The same question appears on vs1063,when codec mp3 audio format and the samplerate is 48khz,the highest frequence response merely reach 16khz, it can not reach 20khz.Can you tell me why ?how to make the frequency response to 20khz,please tell me soon,i'm in a hurry,you would better to measure the chip,if i said is true,please tell me how to do for 20khz,if i said is wrong,maybe the method of measuring is wrong,please tell me how to measure,please!!!

[Henrik]

Both VS1053 and VS1063 are capable of recording audio with a straight frequency response upto 20 kHz. One example application that demonstrates this is the VS1053b WAV PCM Recorder application, which has a frequency response of +- 0.15 dB between 10 - 20000 Hz. The package is available for download at http://www.vlsi.fi/en/support/software/ ... tions.html.

However, Ogg Vorbis and MP3 are both highly compressed, lossy audio formats where upto over 90% of the original audio data bits are removed (example: 48 kHz 16-bit stereo PCM which is 1536 kbit/s is compressed to 128 kbit/s MP3 format). Because they are lossy audio formats, quality always decreases when encoding.

So, to get such high compression, lots of audio data needs to be removed. It is important to remove audio data in such a way that sound quality is kept as good as possible.

Most people cannot really hear frequencies over 15 kHz very well or not at all. Because of that reason, the VS1053 Ogg Vorbis encoder and the VS1063 MP3 encoder remove that data so that there are more bits left for those audio frequencies that are more important. If they tried to encode all frequencies upto 20 kHz, sound quality would actually become worse because valuable bits are used to encode data that cannot be heard anyway! The creators of the MP3 format also thought in a similar way, because they made the MP3 standard in such a way that makes it very inconvenient to encode frequencies above 15 kHz.


About the encoders in VS1053/VS1063.

VS1053:
- Ogg Vorbis: Your observations are correct. The lowest Ogg Vorbis quality setting 0 at 44.1 kHz stereo sets the frequency response upper limit to 13.5 kHz, and the highest setting 5 to 16.5 kHz. As explained earlier, these limits are there to make sound quality better!

VS1063:
- MP3: The highest quality or bitrate settings at 32, 44.1 kHz and 48 kHz (48 kHz recommended) set the frequency upper limit to 15 kHz. This is because of a limitation in the MP3 format itself which is very difficult to circumvent with a real-time encoder.
- Ogg Vorbis: If you record at 44.1 kHz or 48 kHz (48 kHz recommended) and set the quality to higher than 5, your frequency limit will increase. At the highest quality setting 9 the frequency response goes to over 19 kHz.


Conclusion:
Lossy compression is always a combination of compromises. It would be technically possible to create an encoder that would give a frequency response upto 20 kHz even with low bitrates, particularly with Ogg Vorbis. But for real speech and music data it would actually sound worse than an encoder that removes some part of the higher audio spectrum because then there would be less bits left to encode in the more important lower audio spectrum.

In a nutshell:
Don't play the numbers game with compressed formats. Listen to the sound with proper headphones or a good stereo system instead!

[vito1989]

dear henrik,
thanks!!!

[Churmenton]

hi,
it maybe diffcult to make the frequency response reach 20khz when mp3/ogg encoding,but can we develop a patch to make frequency response reach 20khz?without considering the sound quality!for example,using the lowest bitrate to achieve 20 KHZ,
in the VS1053b WAV PCM Recorder application,it has a frequency response of +- 0.15 dB between 10 - 20000 Hz.but if i use audio format IMA ADPCM to encoding,can the frequency response reach 20khz?
please reply me soon

[Churmenton]

dear henrik,
can you reply foregoing mail soon?i'm in a hurry!!!thank you!!!

[Henrik]

it maybe diffcult to make the frequency response reach 20khz when mp3/ogg encoding,but can we develop a patch to make frequency response reach 20khz?without considering the sound quality!for example,using the lowest bitrate to achieve 20 KHZ,

Theoretically this is possible for even low bitrates. However, getting good sound quality in a real-time encoder where we only have 30-50 MIPS of processing power (compared to some 3000 MIPS in PC computers) is very difficult to begin with. After getting the encoders to work at almost the same quality as PC encoders, we cannot release an encoder that would intentionally harm sound quality. Also, for MP3 encoding frequencies over 15.6 kHz requires specific code because the standard makers decided to make the highest frequency band different from others. That makes it impossible to integrate that into the MP3 encoders available in VS10xx ICs.

but if i use audio format IMA ADPCM to encoding,can the frequency response reach 20khz?

If you record IMA ADPCM at 48 kHz, it can encode frequencies up to 20 kHz. However, be warned that when compared to OGG / MP3 files, bitrates will be high (389 kbit/s for stereo 48 kHz) and sound quality will be significantly worse, because IMA ADPCM as a format is designed to be a simple speech codec, not a high-fidelity music encoder. This is a limitation of the IMA ADPCM format, not our implementation.

Kind regards,
- Henrik