For professional quality transcription of audio signals, a floating-point representation can give substantial savings over a straight binary representation. This paper shows, based on psychoacoustic data of masking of noise, how floating-point mantissa length, radix value, and exponent length can be arrived at. Listening tests, using real-time computer-generated music, yielded excellent agreement with the calculated values.
(See document for exact affiliation information.)
https://aes2.org/publications/elibrary-page/?id=3374
(675KB)
Click to purchase paper as a non-member or login as an AES member. If your company or school subscribes to the E-Library then switch to the institutional version. If you are not an AES member Join the AES. If you need to check your member status, login to the Member Portal.
