View Single Post
Old 11-05-2009, 12:35 PM   #1
Big Daddy Big Daddy is offline
Blu-ray Champion
Big Daddy's Avatar
Jan 2008
Southern California
Default Understanding Dialog Normalization


Reference level is defined for film mixing and movie theaters. Every studio and movie theater is calibrated according to this level. It represents an average of 85dB for the regular speakers on the SPL meter (set on C weighting and Slow) using a band limited (500Hz to 2,000Hz) pink noise at the listening position. The peak level is set 20dB higher at 105db and the LFE peak level is set +10dB higher to a maximum of 115dB. The purpose of the +10 dB gain for the LFE channel is to increase the dynamic range of bass sound such as explosions and crashes. This means when the receiver master volume is set to 0dB, the regular speakers are expected to play a peak level of 105dB and the subwoofer is expected to produce a peak output level of 115dB. This is louder than most people can tolerate, so people normally set the master volume much lower than 0 when watching movies or listening to music. Furthermore, such loud bass level places a heavy burden on the subwoofer and requires multiple high-end subwoofers to produce it accurately.

Because 85dBC test tones can be very loud in a small home theater room and can damage hearing, receiver manufacturers through the encouragement by Dolby and THX decided that a reasonable test-tone level is 75dB and that is the level that most receivers use.

To summarize,
  • Reference Level is 1.85v line level = 0dB VU meter = 85db playback level.
  • 105dB Peak level = 0dB (Full Scale).
  • 85dB Average Level = -20dB (Full Scale).
  • 75dB Average Reference Level = -30dB (Full Scale).
  • dBFS (Full Scale) = unit of measure for the amplitude of digital audio signals.
  • The reference level is "0" dBFS, which is also the maximum signal amplitude that can be stored digitally in a typical digital audio recording system.
  • Signals louder than 0dBFS just produce clipping (truncation of the waveform, hence distortion).
When calibrating your audio system, the receiver plays pink noise that is recorded at 75dB (-30dB FS). When the individual speaker levels are set to 75dB at the listening position, as measured by an SPL meter, the effects of speaker sensitivity and room acoustics are accounted for and the speakers are all level-matched against the Reference Level.

You can use either the internal test tones of a receiver or an external disc. The internal test tones of most receivers are band limited and recorded at 75dB level (-30 dB FS). External calibration signals on most discs are typically full-range pink noise and recorded at 85dB (-20dB FS). It really doesn't make a lot of difference which method you use as long as all the speakers are balanced. However, it is important to remember that when you play the internal test tones inside a receiver/processor, they normally bypass all post processing, including equalization.


Metadata is the data sent along with the coded audio signal such as Dolby Digital to describe this signal to the receiver/decoder.

Metadata can contain different types of information such as the 3 Ds.:
  • Dialog Normalization (Level Control): This metadata sets the overall program playback level for loudness normalization (dialognorm).
  • Dynamic Range Control: This metadata has the ability to control the dynamic range of the decoded program by the receiver. The values are calculated in the encoder based on dynamic range profiles selected by the sound engineer. There are five standard profiles: Film Standard, Film Light, Music Standard, Music Light, and Speech.
  • Down-Mixing: Information on how the decoder should perform two-channel or three-channel mix from a 5.1 soundtrack. Metadata for down-mixing tells the player what the relative level of the channels should be in the fabricated Pro Logic mix. Again, this is information for your decoder's benefit, and the sound information itself is not affected by its presence.


Let's think of a soundtrack as a vertical bar. Dynamic Range is the difference between the quietest sound and the loudest sound. Cassette tapes can have a dynamic range of 60 dB, CD audio 80 dB, and Dolby Digital on DVD can have a dynamic range of 105 dB.

Reducing dynamic range in simple terms means raising the level of quiet sounds and lowering the level of loud ones such that there is less of a change. By invoking dynamic range control, you will hear all of the soundtrack but not disturb others with loud peaks.

Source: Home Theater HiFi


The audio level between different TV programs, channels, and in particular commercials is very inconsistent. Dolby Laboratories realized that these radical differences between different types of sources, channels and content would be extremely inconvenient for a TV viewer to continually have to adjust the audio level while watching TV. Because of this, Dolby Labs decided to implement a feature called Dialog Normalization. Dolby includes this feature in metadata of any Dolby Digital bitstream. The only piece of metadata that is mandatory for consumers is dialognorm values. Other example such as downmix metadata is optional and the sound engineers can use it at their discretion.

In a soundtrack such as Dolby Digital, the loudest level is expressed as 0dB and the quietest level as -105dB. The Dialnorm value expresses the level of dialogue as how much lower it is than the peak 0dB. A value of -31 indicates 31dB below the peak. This is the value at which no volume adjustment is performed by the receiver/decoder. A Dialnorm value of -27 would indicate to the receiver/decoder that the dialogue is at a point 27dB below the peak, or 4dB higher than a program with a Dialnorm value of -31. The decoder would then turn the volume down by 4dB. A Dialnorm value of -25 would call for a 6dB reduction and so on. The -27dB setting yields a very natural level for talking and has been the standard level for dialogue in motion picture soundtracks.

Dialogue Normalization works by assigning a relative numerical value as described above to the average dialogue level of a given program, commercial, movie etc. This numerical value should be calculated using hardware or software tools offered by Dolby and is then encoded into the Metadata by the content creator or broadcaster. The graph below shows some of the typical relative levels between different types of source material. The black line and the numerical number next to that line, would be the dialog normalization level.

This numerical value is used by the receiver/decoder to automatically adjust the level of the audio, so that during playback all these different types of program material have the same average dialog level. The graph below shows the resulting adjusted level.

Dialog Normalization does not affect the original audio signals that are fed into the encoder such as the level or the dynamics of the program. The receiver/decoder reads the dialnorm value in the metadata and adjusts the level of audio programs so that the dialog is at a consistent and uniform level. It is important to note that decoders are required to make use of this metadata parameter and apply the proper normalization/attenuation based on the transmitted dialnorm value to the decoded audio program.

  • Dialog Normalization is not the same thing as Dynamic Range Control (DRC).
  • Dialog Normalization does not reduce dynamic range.
  • Dynamic Range Control is the same as Night Mode.
  • Dialog Normalization does not affect the level of dialogue with respect to the other channels or content.
  • Dialog Normalization does not adversely affect the signal-to-noise (S/N) ratio.
  • The default settings for Dolby encoders is -27dBFS.
  • The default dialog setting for DTS is -31dBFS.
  • DTS soundtracks, unlike Dolby Digital, are not attenuated by 4 dB by the receiver/decoder and as a result DTS movies play 4dB louder than Dolby Digital movies.
  • It should be noted that most THX-certified receivers and processors attenuate DTS movies by 4dB.
  • If you want your receiver/processor to report the dialnorm values, you must set the player's audio output to bitstream. In the Denon receivers, this value is reported in the menu under Information, Audio Input Signal. For other receivers, please check the manual or the internal menu of the receiver.


Although the metadata contains different types of information, the only one that is mandatory and must be used is Dialog Normalization. It is important to remember that there are many movies that do not use DN. The dialog normalization vlaues can be between zero (no attenuation) to -31 (31 dB attenuation). The most common value for the dialog normalization is -4 dB.

The attenuation due to Dialog Normalization is performed in the digital domain before the D/A conversion. The PCM output that is outputted from the player is transcoded after the values in the metadata are applied. When DTS-HD MA and Dolby TrueHD are transcoded to PCM, all metadata (DN, etc.) is then stripped away from the encoded PCM stream. This happens because uncompressed LPCM cannot carry a secondary stream like metadata, and this is the reason why receivers do not show any DN offset.

Digital attenuation takes place by multiplying each successive word in the input data stream by a coefficient that is less than one. For example, if the incoming stream is multiplied by a coefficient of 0.5, the output will be reduced by half. This is equivalent to a volume reduction of 6 dB. In a binary system, a reduction of 6 dBs is represented as shifting the signal from the most significant bit towards the least-significant bit by one bit. This will result in a reduced resolution by one bit. This is illustrated in the following diagram:

All Diagrams are created by Big Daddy

Using the same analysis, if the incoming signal is attenuated by a factor of eight, this will result in a volume reduction of 18 dB and the resolution will be lost by 3 bits.

Now, let’s apply the above analysis to a practical 16 bit audio from a blu-ray movie. In practice, 24-bit or 32-bit processors and digital/analog converters are used. As a result, a 16-bit signal from a BD can be attenuated without loss of the original 16-bit information.

Similarly, you can find an example of 24-bit digital volume control attenuation in the following. Let’s assume a 16-bit input from a BD or DVD fed into it. The 24-bit DSP performs digital filtering and resolution enhancement, producing a 24-bit signal. The additional bits are represented by the dotted lines being fed to at the eight least-significant bits of the 24-bit DAC. Let’s assume the original 16-bit audio input is attenuated by 18 dB. This results in the loss of the three least significant bits, but the original 16-bit information from the BD/DVD is preserved.

Even if the 16-bit signal is attenuated by 30 dB, as shown below, there will be a reduction of five bits, but the original 16-bit signal is preserved intact.

The Oppo BDP-95 and BDP-105 players use 32 bit processing. The digital volume controller on these units affects only the analog signals coming out of the player. I have seen concerns expressed by quite a few people over the loss of resolution if the digital volume attenuation is used. Let me assure you that this concern is completely unwarranted. If a 24 bit signal is fed into a 32 bit processor, the digital volume controller can be attenuated up to 48 dBs (8 bits) without any loss of resolution. If the input signal is from a CD (16 bits), the volume can be attenuated up to 96 dBs (16 bits) without any loss of resolution.

Strictly speaking, digital attenuation does reduce resolution. However, analog volume controls and associated circuitry have many negative side effects such a loss of resolution caused by analog circuitry and connectors in the signal path. The degradation and noise of the analog circuitry are present under all conditions, regardless of playback volume level.

Dialogue Normalization: Friend or Foe
Why are TV commercials so loud? : Audio Design Labs Inc.
Managing DTV LOUDNESS with dialnorm
nodef Home Theater: 0 dB THX reference level

Last edited by Big Daddy; 02-18-2013 at 11:37 PM.
  Reply With Quote
Thanks given by:
ChromeJob (07-15-2015)