Research Report LGRP#1

On this page:

• Significance of the study 
• Abstract of the study
• Overview of the Methodology, Main Results and Conclusions.
    - Blind and non-blind testing by guitarist and listener (3 sessions)
    - Blind and non-blind testing by public audience (2 sessions)
• Additional information: chard with the woods used for the 10 non-tropical guitars. 
• Downloadable PDF of the study


For over a century it has been widely believed that high quality classical guitars – their soundboards being the major exception – have to be constructed from tropical woods. This applies particularly to backs and sides (rosewood/mahogany) but also to necks (mahogany/Spanish cedar), fingerboards and bridges (ebony/rosewood). Indeed most builders and players alike consider such tropical hardwoods as the preferred, if not the only choice for instruments of the highest tonal quality.

Whilst some non-tropical back and side woods have become established for certain applications - most notably cypress for flamenco guitars and maple for the more affordable “student” classical models - experimentation with other, alternative non-tropical woods has only ever occurred on a small scale and has not delivered the empirical proof required to indicate that such woods can be regarded as acceptable tonewoods. Until this present study, there has been no reliable, scientific investigation into the suitability of non-tropical species such as alder, ash, oak, plane, birch, false acacia, chestnut etc. for classical guitar building.

In order for the widespread belief in the tonal superiority of tropical wood in guitar making to be validated, research would be required to prove that, under blind conditions, experienced guitarists and listeners show a clear preference for guitars made from tropical woods, and are able to reliably distinguish them from their non-tropical counterparts.

This study, however, shows that first-rate guitar players and experienced listeners alike are unable either to register a clear preference between the two sets of guitars or to distinguish between both wood groups at anything better than chance levels.

Furthermore, under non-blind conditions, the non-tropical wood guitars saw, on average, a marked fallback in preference of some 50%. This would indicate that sound perception is strongly influenced by visually transmitted information such as the aesthetic qualities of an instrument, or the preconceptions surrounding good and bad tonewoods that their recognition allows.


Research Objective:
A comparative study of the sound preferences between guitars made from
tropical woods, and guitars made from non-tropical woods.

Organized tests:
Blind and Non-Blind playing and listening tests by
1)  Guitarist/Listener (3 sessions)
2)  Public Audience (2 sessions)

Tested guitars:
- 10 classical  guitars made from non-tropical woods
- 5 classical  guitars made from tropical woods 

All guitars of the same model: Torres FE19
- All guitars of the same high quality standard
- All tops made out of european spruce of the 
   same quality and the same bracing pattern.

Testing period:
March > July  2014

All blind tests show that 
guitars made from non-tropical wood species were equally preferred for sound quality as those made from tropical wood.
Non-blind tests show a strong fallback in sound 
appreciation experienced by non tropical wood guitars.

The need for the use of tropical wood in acoustic and classical guitar building seems to be a myth because
blind tests have clearly indicated that non-tropical woods can be used to make guitars whose tonal quality is equal to that of their tropical counterparts.
The fallback in sound appreciation experienced by
non-tropical wood guitars in the non-blind tests, strongly suggests that preference is heavily influenced by
preconceptions about what guitars should look like and about what exactly constitutes a tonewood.

• OVERVIEW of the methodology, main results and main conclusions of blind and non-blind
   playing and listening tests by
Guitarists/Listeners and (2) Public Audiences

• (1) BLIND and NON-BLIND TESTS by Guitarist and Listener (3 sessions)
        The test-sessions consisted:
           - Pair testing, Blind
           - Group testing, Blind
           - Assessment of aesthetic, Non-blind
           - Pair testing, Non-blind
           - Interview with the guitarist and listener

Abbreviations: G = Guitarist / L = Listener / T = Tropical / NT = Non-Tropical


- 12 pairs were tested.
- Each pair consisted of a Tropical (T) and a Non-Tropical (NT) guitar.
- 2 pairs were presented twice to test consistency of preferences / actually there  were only 10 pairs
   but the guitarist (G) and the listener (L) were unaware of this.
- G was blindfolded. L was behind an acoustically transparent screen.
- G played the same piece of test music on every guitar.
- G played each of the two guitars (1 T and 1 NT in a pair) in succession, time Limit: 30 seconds per guitar.
- G and L are then asked which of the 2 guitars they prefer. Three possibilities: No. 1 / No. 2 / No Preference. 

- 4 groups of 5 guitars were presented / Each group consisted of 3 or 4 Non-Tropicals (NT’s) and 1 or 2 Tropicals (T’s)
- In each group 1 guitar was presented twice to test voting consistency (so actually only 4 different guitars were
  per group).
- G was blindfolded. 
- Assessment method per group: To get a first impression of the guitars in the group, G played each guitar for 30
After that G was now free to play what he wanted and for as long he wanted. He could ask for any guitar in any
- Finally G rated the guitars:  Very good / Good / Average / Poor / Very poor,  and he was asked to say which guitar he found
   ‘the best’ in the group 
(he was allowed to give the same rating to more  then 1 guitar if he wanted).
   Finally G was asked to explain, in his own words, why he found his favorite guitar the best out of the group.
Note: Group testing blind was only conducted with the Guitarists, not with the Listeners

- Same as ‘Pair Test Blind’, but now the 10 pairs were assessed without any pairs being tested twice.
- The order of the pairs was different but the order of the guitars within the pairs was the same as in ‘PairTesting Blind'



 Fig 1

 Fig 2

RESULTS & CONCLUSIONS of the BLIND and NON-BLIND TESTS by guitarists and listeners over 3 sessions

There was a strong consistency in the results per guitarist and listener across all of the three sessions.
The tests were consistently good in quality and performance and therefore provide a reliable source of information.

Score consistency for double pairs and double guitars in the groups is much better for guitarists than for the listeners
(pair testing blind: average guitarists scores: 83 %, average listeners scores: 17 %   (Fig 2)
Playing the guitars allows for a much better consistency in assessment than just ‘passive’ listening. Results of (playing) guitarists are therefore more reliable (although preference results between guitarists and listeners are very similar).

Non-Tropical Guitars and Tropical Guitars were equally preferred for sound quality (around 50/50%). Fig 1
The only exception was Listener 1 in the pair test, but his consistency in this test was 0%).
Experienced guitar players and listeners were unable to distinguish T’s from NT’s at better than chance levels. All blind tests show that it is possible to make equally good sounding guitars from both Tropical and Non-Tropical Woods.

When guitarists and listeners could see the guitars, and they knew from which woods they were made, we see a very strong difference in sound perception: ± 50% fallback for the Non-Tropicals compared with the same Non-Tropicals in the ‘blind’ testing (from 50% blind to 25% non-blind, Fig 1)
Sound perception is strongly influenced by aesthetics and preconceptions of what are the ‘best’ woods for making guitars. Prejudices play an important role in what guitar players and listeners think sounds good or bad.

• (2)  BLIND and NON-BLIND TESTING by public audience in Belgium and Finland


- Two audience test sessions were carried out in concert hall conditions, 
   one in Belgium (44 listeners),  one in Finland (22 listeners).
- A session consisted 
Blind and Non-Blind pair tests.
- The same guitars were used as in the 'Blind' and 'Non-Blind pair tests for Guitarists/Listeners (see above)
-  A guitarist played the guitars in pairs  (one T and 1 NT) to a ‘Blind’ audience  (screened off from the guitarist)
   and subsequently to a Non-Blind audience (that could see the guitars and were told the woods used for backs and sides).
- The audience was asked to vote for their preferred guitar in each pair.
   Three preference possibilities:  No. 1 / No. 2 / No Preference.
- In addition they had to note if they found the sound of the 2  guitars in one pair; Very close/Close/Different/ Very Different.
- In the Audience Blind testing 2 pairs were presented twice (as in the Pair Test Blind, see above)  to check consistency
   of voting preference.



RESULTS & CONCLUSIONS of the public audience tests in Belgium and Finland

- In the Audience Blind Tests NT guitars and T Guitars were equally rated for sound quality  (around 50/50% ) (Fig 3).
- Consistency in voting preferences was low. (Fig 5 )
- The Closeness/Difference rates (Fig 6) tend to be greater in the 'Very Close/Close' zone than in the 'Different/Very Different'

-  It was very difficult to distinguish between tropical and non-tropical guitars under blind audience conditions.
    Listeners were unable to distinguish T’s from NT’s at better than chance levels.
-  Non-tropical woods can be used to make guitars of equal sound quality to those made with tropical woods.

- We see less difference between the Blind and Non-Blind audience tests, ( Fig 3 and 4)  than we did between
   the 'Blind' and 'Non-Blind Pair Tests' for guitarists/listeners (Fig 1). 
   This is most likely due to the fact  that the audience (particularly in Belgium) had a significant number of
   “Non-Tropical“ guitar builders and supporters, and was consequently less biased towards tropical woods.

• Additional information:
   Wood species used for the 10 Non-Tropical guitars


Questions and remarks about this study can be send to

• The overview of the study can be downloaded  (PDF 1,1 MB)