DVDfile.com forum for DVD, Blu-Ray, and HD-DVD - View Single Post

Tyler Durden · 11-03-2000, 10:24 AM

Hello All

Many of you may well be aware of this new software from THX that is encoded on many of the discs they slap their name on. The first to feature this new system was, of course, Fight Club (and what a brilliant film it is too.) However, I would advise against anyone using these test tones to calibrate their system as the tones are incorrect - some by as much as 10db. Use the tones to check the polarity of your system and that the corect channles have been assigned to their proper place by all means but under no circumstances should you use them to set the SPL level for your speakers. If you do then your system settings will be totally incorrect. I know that many of you are paranoid about what you read on the net and lest I be accused of posting misinformation on this forum, I shall offer you the following info. which should keep all you, "doubting Thomas's" at bay.

The THX Optimode on-screen instructions state that the output levels achieved in a correctly calibrated system when using Optimode should be 75dB SPL. This is the same volume achieved when calibrating using the internal test noise of a THX receiver/processor or any Dolby Digital hardware whose volume control operates in relation to 'reference' level.

For test noise to produce 75dB SPL, it must be recorded at -30dBFS (that's thirty decibels below full-scale, the maximum output of a digital system). As Dolby Digital's maximum output is 105dB for all five main channels, -30dBFS equates to 75dB SPL.

Unfortunately, with 'Fight Club' being the only exception, the THX Optimode calibration noise has been recorded at -20dBFS (twenty decibels below full-scale) which results in an SPL output of 85dB, 10dB above what the on-screen graphics state should be the case.

THX assure me that they will correct the amplitude of all future THX Optimode signals to bring them into line with the 'standardised' calibration output level of 75dB SPL.

The users of some processors/receivers have not been reporting 10dB discrepancies however, 6dB is often the figure quoted. This can be explained by examining how their decoders may be reacting to the Dolby Digital bitstream used to convey the Optimode calibration signals.

The Dialog Normalization Factor

Within all Dolby Digital bitstreams (carried by the SI header of each frame) is a data instruction called 'dialog normalization' — sometimes known as the 'reference offset' value. As the consumer version of Dolby Digital was originally designed for HDTV applications, the dialog normalization function was included so that broadcasters could control the relative volume level of programme material, especially when different material was shown back-to-back. Don't let the name fool you, the entire soundtrack is affected, not just dialog.

Dialog normalization instructs the Dolby Digital decoder to reduce the level of its outputs by the amount specified within the bitstream. The typical amount of attenuation applied is 4dB, the 'standard' for Dolby Digital motion picture material. This figure can vary however, for added headroom 'Air Force One' (an isolated example) has a -0dB dialog normalization value while mono material, such as the re-releases of old James Bond movies can have up to -9dB dialog normalization.

As mentioned earlier, the maximum output from the five main channels of a Dolby Digital bitstream (in a calibrated system) is 105dB SLP, but if that bitstream contains a -4dB dialog normalization value, the outputs will be attenuated by the decoder resulting in a maximum output of 101dB SPL.

When producing calibration software, it is therefore vital that the volume level of the calibration noise and the bitstream dialog normalization value are considered as a whole. If calibration noise is recorded at -30dBFS, for it to produce 75dB SPL from a standard Dolby Digital decoder the dialog normalization must be -0dB so that the outputs of the decoder are not attenuated. All of Dolby's own calibration material has a -0dB dialog normalization value.

The THX Optimode calibration signals have a dialog normalization value of -4dB and as is clear, this reduces the output of the decoder and leads to the 6dB discrepancy. This is because we begin with a signal that is 10dB too loud (85dB SPL instead of 75dB) and there is 4dB of decoder attenuation. +10dB + -4dB = 6dB. Therefore, Dolby Digital decoders will correctly produce 81dB SPL when evaluated using the THX Optimode signals. The decoders are not to blame, at fault are the THX Optimode signals themselves.

There are situations in which the Optimode signals do produce the correct results (albeit with a 10dB discrepancy), and this we'll tackle later in this post.

Artificial Level Adjustments

In recent years, DTS has become more popular and it was inevitable that DTS soundtracks would be compared to their Dolby Digital counterparts (ad infinitum). No matter which you prefer, one thing is clear: DTS soundtracks are almost always louder than Dolby Digital soundtracks. This is because the DTS bitstream does not carry an equivalent of Dolby Digital's dialog normalization bit and as we have learnt, in the majority of cases this instructs the Dolby Digital decoder to apply 4dB of attenuation to its outputs.

In an effort to standardise the output levels of both Dolby Digital and DTS, some THX receiver and controller manufacturers (Denon being one example) are introducing a global 4dB gain to their Dolby Digital decoders, perhaps under instruction from the THX department of LucasFilm. This directly counters the 'standard' -4dB Dolby Digital dialog normalization bit and results in matched output volume levels between Dolby Digital and DTS source material. As manufacturers can reduce their internal calibration noise (against reference) by the same amount, output levels remain at the correct level unless the Dolby Digital signal carries a -0dB dialog normalization bit, in which case the outputs will have an artificial gain of 4dB.

It would appear that the THX Optimode calibration noise has been recorded with the small number of decoders that employ this 4dB boost/cut scheme in mind. This is why some users — typically those with recently introduced hardware — report that the Optimode signals produce predictable output levels, even though as we have discussed, with the exception of 'Fight Club', they will be 10dB too loud.

Conclusion

The moral of this story is clear. Heed the warnings from THX and do not use the Optimode calibration signals to evaluate the output levels of your system... unless all you want is to differentiate one channel from another. Not only are the signals recorded at -20dBFS in error, but based on my own experiments using a number of THX and non-THX receivers and controllers, are also adversely affected by the use of a -4dB dialog normalization bit, contrary to the standards employed by Dolby Laboratories.

As an alternative, I would recommend 'Video Essentials'

Hope that helps and gives some insight into the wonders of HT.

[This message has been edited by Tyler Durden (edited 11-03-2000).]

11-03-2000, 10:24 AM	#1 (permalink)
Tyler Durden Actor Join Date: Jun 2000 Location: London	Beware of THX Optimode Hello All Many of you may well be aware of this new software from THX that is encoded on many of the discs they slap their name on. The first to feature this new system was, of course, Fight Club (and what a brilliant film it is too.) However, I would advise against anyone using these test tones to calibrate their system as the tones are incorrect - some by as much as 10db. Use the tones to check the polarity of your system and that the corect channles have been assigned to their proper place by all means but under no circumstances should you use them to set the SPL level for your speakers. If you do then your system settings will be totally incorrect. I know that many of you are paranoid about what you read on the net and lest I be accused of posting misinformation on this forum, I shall offer you the following info. which should keep all you, "doubting Thomas's" at bay. The THX Optimode on-screen instructions state that the output levels achieved in a correctly calibrated system when using Optimode should be 75dB SPL. This is the same volume achieved when calibrating using the internal test noise of a THX receiver/processor or any Dolby Digital hardware whose volume control operates in relation to 'reference' level. For test noise to produce 75dB SPL, it must be recorded at -30dBFS (that's thirty decibels below full-scale, the maximum output of a digital system). As Dolby Digital's maximum output is 105dB for all five main channels, -30dBFS equates to 75dB SPL. Unfortunately, with 'Fight Club' being the only exception, the THX Optimode calibration noise has been recorded at -20dBFS (twenty decibels below full-scale) which results in an SPL output of 85dB, 10dB above what the on-screen graphics state should be the case. THX assure me that they will correct the amplitude of all future THX Optimode signals to bring them into line with the 'standardised' calibration output level of 75dB SPL. The users of some processors/receivers have not been reporting 10dB discrepancies however, 6dB is often the figure quoted. This can be explained by examining how their decoders may be reacting to the Dolby Digital bitstream used to convey the Optimode calibration signals. The Dialog Normalization Factor Within all Dolby Digital bitstreams (carried by the SI header of each frame) is a data instruction called 'dialog normalization' — sometimes known as the 'reference offset' value. As the consumer version of Dolby Digital was originally designed for HDTV applications, the dialog normalization function was included so that broadcasters could control the relative volume level of programme material, especially when different material was shown back-to-back. Don't let the name fool you, the entire soundtrack is affected, not just dialog. Dialog normalization instructs the Dolby Digital decoder to reduce the level of its outputs by the amount specified within the bitstream. The typical amount of attenuation applied is 4dB, the 'standard' for Dolby Digital motion picture material. This figure can vary however, for added headroom 'Air Force One' (an isolated example) has a -0dB dialog normalization value while mono material, such as the re-releases of old James Bond movies can have up to -9dB dialog normalization. As mentioned earlier, the maximum output from the five main channels of a Dolby Digital bitstream (in a calibrated system) is 105dB SLP, but if that bitstream contains a -4dB dialog normalization value, the outputs will be attenuated by the decoder resulting in a maximum output of 101dB SPL. When producing calibration software, it is therefore vital that the volume level of the calibration noise and the bitstream dialog normalization value are considered as a whole. If calibration noise is recorded at -30dBFS, for it to produce 75dB SPL from a standard Dolby Digital decoder the dialog normalization must be -0dB so that the outputs of the decoder are not attenuated. All of Dolby's own calibration material has a -0dB dialog normalization value. The THX Optimode calibration signals have a dialog normalization value of -4dB and as is clear, this reduces the output of the decoder and leads to the 6dB discrepancy. This is because we begin with a signal that is 10dB too loud (85dB SPL instead of 75dB) and there is 4dB of decoder attenuation. +10dB + -4dB = 6dB. Therefore, Dolby Digital decoders will correctly produce 81dB SPL when evaluated using the THX Optimode signals. The decoders are not to blame, at fault are the THX Optimode signals themselves. There are situations in which the Optimode signals do produce the correct results (albeit with a 10dB discrepancy), and this we'll tackle later in this post. Artificial Level Adjustments In recent years, DTS has become more popular and it was inevitable that DTS soundtracks would be compared to their Dolby Digital counterparts (ad infinitum). No matter which you prefer, one thing is clear: DTS soundtracks are almost always louder than Dolby Digital soundtracks. This is because the DTS bitstream does not carry an equivalent of Dolby Digital's dialog normalization bit and as we have learnt, in the majority of cases this instructs the Dolby Digital decoder to apply 4dB of attenuation to its outputs. In an effort to standardise the output levels of both Dolby Digital and DTS, some THX receiver and controller manufacturers (Denon being one example) are introducing a global 4dB gain to their Dolby Digital decoders, perhaps under instruction from the THX department of LucasFilm. This directly counters the 'standard' -4dB Dolby Digital dialog normalization bit and results in matched output volume levels between Dolby Digital and DTS source material. As manufacturers can reduce their internal calibration noise (against reference) by the same amount, output levels remain at the correct level unless the Dolby Digital signal carries a -0dB dialog normalization bit, in which case the outputs will have an artificial gain of 4dB. It would appear that the THX Optimode calibration noise has been recorded with the small number of decoders that employ this 4dB boost/cut scheme in mind. This is why some users — typically those with recently introduced hardware — report that the Optimode signals produce predictable output levels, even though as we have discussed, with the exception of 'Fight Club', they will be 10dB too loud. Conclusion The moral of this story is clear. Heed the warnings from THX and do not use the Optimode calibration signals to evaluate the output levels of your system... unless all you want is to differentiate one channel from another. Not only are the signals recorded at -20dBFS in error, but based on my own experiments using a number of THX and non-THX receivers and controllers, are also adversely affected by the use of a -4dB dialog normalization bit, contrary to the standards employed by Dolby Laboratories. As an alternative, I would recommend 'Video Essentials' Hope that helps and gives some insight into the wonders of HT. [This message has been edited by Tyler Durden (edited 11-03-2000).]