I'm designing and measuring some rooms and most of all I use EBU 3276 recommendations. An important part is that about reflections and the criterion is they should be under -10 dB compared to direct sound.
The diagram from Jonathan Sheaffer's thesis you included in your post shows detection thresholds from experiments in anechoic rooms with a single loudspeaker as direct sound source and a single reflection. Experiments performed using a complete sound field with all reflections and reverberation obtain different thresholds, for speech for instance for a single 10 ms lateral reflection it is increased by 4 dB (Olive 1989) or 6 dB (Bech 1995).
Detection thresholds for music depend on the music motive, and are between – 25 and -18 dB re: direct sound, 10 ms single lateral reflection, single speaker for direct sound, anechoic room (Schubert 1966). If detection thresholds for music change in a similar way as they do for speech the figure for a complete sound field would be between about -20 and – 13 dB re: direct sound, which means that for lateral reflections a level of 10 dB below direct sound is probably not enough.
Further one should also take into account the the natural level of reflections and determine whether or not it is below the detection threshold, which would be good. These levels depend on the directivity characteristics of the loudspeaker, the absorption coefficients of the room surfaces, and the attenuation due to distance.
Olive et al., “The detection of reflections in typical rooms”, J. of the Audio Engineering Society 1989, S.539
Bech, “Timbral aspects of reproduced sound in small rooms I”, J. of the Acoustical Society of America 1995, vol.97, no. 3, S.1717
Schubert, “Detectability of single reflections for music” (Untersuchungen über die Wahrnehmbarkeit von Einzelrückwürfen bei Musik), Technische Mitteilungen RFZ 1966, vol. 10, no. 3, p.124
EBU says something like the decays should be even to avoid echoes. But how much? There are echo criteria for big rooms like Dietsch & Kraak's but they don't work for small ones because they require a long delay.
Echo threshold are different for different signals, and are about 1-5 ms (clicks), 10-20 ms (noise), 30-50 ms (speech), 80-150 ms (Blauert 2005, Dietsch 1986, Litovsky 1999).
First reflections in small rooms come at delays of under 20 ms (Devantier 2002), so except for sounds like clicks there is no problem, the precedence effect will do its job.
Blauert et al. (2005), “Acoustical communication: The precedence effect”, Proceedings FORUM ACUSTICUM BUDAPEST, OPAKFI Budapest
Devantier (2002), “Characterizing the amplitude response of loudspeaker systems”, Audio Engineering Society preprint 5638
Dietsch et al. (1986), “An objective criterion for the measurement of echo disturbance during presentation of music and speech” (Ein objektives Kriterium zur Erfassung von Echostörungen bei Musik- und Sprachdarbietungen), Acustica, vol. 60, p.205
Litovsky et al., “The precedence effect”, J. of the Acoustical Society of America 1999, vol. 106, no. 4, pt. 1, p.1633
EBU 3276 actually says: “The reverberation field should be sufficiently diffuse in the listening area to avoid perceptible acoustic effects such as flutter echoes.”
The condition for the generation of flutter echoes is that the duration of the impulse is much shorter than the distance between the reflectors divided by the speed of sound in air:
t0 << l/c
If the duration is longer no flutter echo will be generated but one of the room modes will be excited.
Maa, “The flutter echoes”, J. of Acoustical Society of America 1941, vol. 13, p.170
Klaus