Change search
ReferencesLink to record
Permanent link

Direct link
Restore Sound Quality of Old Records, Using Statistical Estimation Based on Several Copies of the Same Original Recording
KTH, School of Electrical Engineering (EES), Sound and Image Processing.
2012 (English)Independent thesis Advanced level (professional degree), 20 credits / 30 HE creditsStudent thesis
Abstract [en]

The purpose of this project is to create and implement digital signal processing methods to improve the sound quality of pre-1940s gramophone records by using several flawed (scratched, worn by use and aged) copies of the same original recording. Current restoration techniques are often based on improving the best available copy. This can pose a problem when there is no record of good quality to be found. There are many different ways of implementing filters to improve the sound quality. One method which gives a big improvement in noise reduction can be made by using the left and right channels of a single gramophone disc. This can be done because old gramophone records were cut monophonic which results in two channels with the same music signal but affected differently by noise from for example cracks and dust. When filtering several recordings a problem faced is that each record is poorly synchronized in the time-domain because of fluctuations in turntable speed, sampling rate and from disc deformities. Three different methods of filtering the time-aligned multiple records are created and tested. Testing shows that common gramophone 'click', and 'pop' noises can be fully suppressed and that quite significant signal-to-noise ratio gains can be expected depending on method used and the quality and quantity of gramophone records.

Place, publisher, year, edition, pages
2012. , 35 p.
EES Examensarbete / Master Thesis, Xr-EE-SIP 2012:003
National Category
Electrical Engineering, Electronic Engineering, Information Engineering
URN: urn:nbn:se:kth:diva-98755OAI: diva2:538858
Educational program
Master of Science in Engineering - Electrical Engineering
2012-06-28, Conference room at SiP, Osquldas väg 10, Stockholm, 14:00 (English)
Available from: 2012-07-16 Created: 2012-07-02 Last updated: 2012-09-10Bibliographically approved

Open Access in DiVA

No full text

Search in DiVA

By author/editor
Lindström, Per
By organisation
Sound and Image Processing
Electrical Engineering, Electronic Engineering, Information Engineering

Search outside of DiVA

GoogleGoogle Scholar
The number of downloads is the sum of all downloads of full texts. It may include eg previous versions that are now no longer available

Total: 53 hits
ReferencesLink to record
Permanent link

Direct link