HK1053534B - Method and apparatus for enhancing source coding and decoding by adaptive noise-floor addition and noise substitution limiting - Google Patents

Description

Method and system for enhancing source coding and decoding using adaptive noise-floor addition and noise substitution limiting

Technical Field

The present invention relates to a source encoding system using High Frequency Reconstruction (HFR) such as spectral band replication SBR [ WO98/57436] or related methods. It improves the performance of both the high quality process (SBR) and the low quality replication process [ u.s.pat.5,127,054 ]. It is applicable to speech coding systems and natural audio coding systems. Furthermore, with adaptive noise floor addition, the present invention can be advantageously used in conjunction with natural audio codecs with or without high frequency reconstruction to reduce the audible effects of band-off that typically occur under low bit rate conditions.

Background

The presence of random signal components is an important property of many musical instruments and human voices. If the perceived signal is a natural sound, it is important to reproduce these noise components, which are often mixed with other signal components. In high frequency reconstruction, under certain conditions, noise must be added to the reconstructed high frequency band in order to obtain a noise content similar to that in the original signal. This necessity arises from the fact that most harmonic sounds produced by, for example, a reed or bow-string instrument have relatively high noise levels in the high frequency region, as compared to the noise levels in the low frequency region. In addition, the emitted harmonic sound sometimes contains high-frequency noise, so that there is no similarity between the high-frequency band noise level and the low-frequency band noise level in the signal. In either case, frequency substitution, i.e., high quality SBR, as well as any low quality replication process, sometimes suffers from lack of noise in the high frequency band of the replication. Even more, the high frequency reconstruction process often includes some sort of envelope adjustment, where it is desirable to avoid unwanted noise replacing harmonics. It is therefore important to be able to increase and control the noise level during high frequency regeneration in the decoder.

At low bit rate conditions, natural audio codecs typically show severe band closure. This is done on a frame-to-frame basis, resulting in spectral holes occurring in an arbitrary manner throughout the encoded frequency range. This situation can cause audible artifacts. This effect can be mitigated by using an adaptive noise floor addition method.

Some prior art coding systems include means for reconstructing the noise component in the decoder. This may allow the encoder to omit the noise component during encoding, thus making it more efficient. However, for this method to be successful, the noise rejected by the encoder during encoding must not contain other signal components. Such hard-decision based noise coding schemes result in relatively low duty cycles since most noise components are mixed in time and/or frequency with other signal components. Moreover, such a solution does not in any way solve the problem of insufficient noise content in the reconstructed high frequency band.

Disclosure of Invention

The invention solves the problem of insufficient noise content in a regenerated high frequency band and a frequency spectrum hole caused by frequency band closing under the condition of low bit rate by utilizing the self-adaptive noise floor increase. It also avoids substitution of harmonics with unwanted noise. This is done by estimating the noise floor level in the encoder, and adaptive noise floor addition and no noise substitution limitation in the decoder.

The present invention provides, in one aspect, a method for enhancing a source coding system that generates an encoded signal by encoding an original signal, the original signal having a low-band portion and a high-band portion, the encoded signal including the low-band portion of the original signal and not including the high-band portion of the original signal, the method comprising the steps of: estimating a noise floor level of a highband portion of the original signal, the noise floor level being a measure of a difference between a first spectral envelope determined from each local minimum in a spectral representation of the original signal and a second spectral envelope determined from each local maximum in the spectral representation of the original signal; and multiplexing the encoded signals including the noise floor level in the low-band portion and the high-band portion of the original signal to obtain an output signal of the encoder.

Another aspect of the present invention provides an apparatus for enhancing a source encoder that generates an encoded signal by encoding an original signal, the original signal having a low-band portion and a high-band portion, the encoded signal including the low-band portion of the original signal and not including the high-band portion of the original signal, the apparatus comprising: an estimator for estimating a noise floor level of the original signal, the noise floor level being a measure of a difference between a first spectral envelope determined by local maxima in a spectral representation of the original signal and a second spectral envelope determined by local maxima in the spectral representation of the original signal; and a multiplexer for multiplexing the encoded signals including the noise floor levels of the low-band portion and the high-band portion of the original signal.

Yet another aspect of the present invention provides an apparatus for enhancing a source decoder that generates a decoded signal by decoding an encoded signal obtained by source encoding an original signal, the original signal having a low-band portion and a high-band portion, the encoded signal including the low-band portion of the original signal and not including the high-band portion of the original signal, wherein the decoded signal is used for high-frequency reconstruction to obtain a high-frequency reconstructed signal including the high-band portion reconstructed from the original signal, the apparatus comprising: a demultiplexer for demultiplexing an input signal comprising the encoded signal and a noise floor level of a highband portion of the original signal, the noise floor level being a measure of a difference between a first spectral envelope determined by each local minimum in a spectral representation of the original signal and a second spectral envelope determined by each local maximum in the spectral representation of the original signal; means for obtaining a representation of a spectral envelope of a highband portion of an original signal; a shaper, which shapes the spectrum of the random noise signal according to the spectral envelope representation of the high-band part of the original signal to obtain a spectrally shaped random noise signal; a regulator for regulating the spectrally shaped random noise signal according to the noise floor level to obtain a regulated spectrally shaped random noise signal; and an adder for adding the adjusted spectrally shaped random noise signal to the high frequency reconstructed signal to obtain an enhanced high frequency reconstructed signal.

Yet another aspect of the present invention provides a method for enhancing a source decoding system that generates a decoded signal by decoding an encoded signal obtained by source encoding an original signal, the original signal having a low-band portion and a high-band portion, the encoded signal including the low-band portion of the original signal and not including the high-band portion of the original signal, wherein the decoded signal is used for high-frequency reconstruction to obtain a high-frequency reconstructed signal including the high-band portion reconstructed from the original signal, the method comprising the steps of: demultiplexing an input signal comprising the encoded signal and a noise floor level of a highband portion of the original signal, the noise floor level being a measure of a difference between a first spectral envelope determined by each local minimum in a spectral representation of the original signal and a second spectral envelope determined by each local maximum in the spectral representation of the original signal; obtaining a spectral envelope representation of a high-band portion of an original signal; shaping the spectrum of the random noise signal according to the spectral envelope representation of the high-band part of the original signal to obtain a spectrally shaped random noise signal; adjusting the spectrally shaped random noise signal according to the noise floor level to obtain an adjusted spectrally shaped random noise signal; and adding the adjusted spectrally shaped random noise signal to the high frequency reconstructed signal to obtain an enhanced high frequency reconstructed signal.

The adaptive noise floor addition and noise substitution limiting method comprises the following steps:

-estimating, in the encoder, a noise floor level of the original signal using a dip-follow (dip-follow) and a peak-follow (peak-follow) added to a spectral representation of the original signal;

-in the encoder, the noise floor level is transformed to several frequency bands, or represented by LPC or any other polynomial;

-smoothing the noise floor level in time and/or frequency in an encoder or decoder;

-shaping, in the decoder, the random noise according to the spectral envelope representation of the original signal and adjusting the noise according to the noise floor level estimated in the encoder;

-smoothing the noise level in time and/or frequency in a decoder;

-adding the noise level to the high frequency reconstructed signal in the regenerated high frequency band or in the shut down frequency band.

-adjusting, in the decoder, the spectral envelope of the high frequency reconstructed signal with envelope adjustment amplification factor limitation.

In the decoder, the frequency resolution is increased by means of interpolation of the received spectral envelope, thus improving the performance of the slicer.

-applying, in the decoder, a smoothing operation to the envelope adjustment amplification factor.

-generating, in the decoder, a high frequency reconstructed signal, which is a sum of several high frequency reconstructed signals originating from different low band frequency ranges, and analyzing the low band to provide control data to this sum.

Drawings

The invention will now be described by way of several illustrative examples without limiting the scope or spirit of the invention, with reference to the accompanying drawings, in which:

FIG. 1 shows a peak trace and a valley trace added to high and medium resolution spectra, and a noise floor to frequency band transform, in accordance with the present invention;

FIG. 2 illustrates a noise floor smoothed in time and frequency in accordance with the present invention;

FIG. 3 shows a spectrum of an original input signal;

FIG. 4 shows the output signal spectrum of the SBR process without adaptive noise floor addition;

FIG. 5 shows the output signal spectrum with SBR and adaptive noise floor addition in accordance with the invention;

fig. 6 shows the amplification factors of a spectral envelope adjustment filter bank according to the invention;

fig. 7 shows smooth amplification factors in a spectral envelope adjustment filter bank according to the invention;

fig. 8 shows a possible embodiment of the invention at the encoder side in a source coding system;

fig. 9 shows a possible implementation of the invention at the decoder side in a source coding system.

Detailed Description

The embodiments described below merely illustrate the principles of the invention for improving a high frequency reconstruction system. It should be understood that various changes and modifications in the arrangement and details described herein will be apparent to others skilled in the art. The invention is, therefore, limited only by the scope of the appended claims, and not by the specific details of the embodiments described and illustrated.

Noise floor level estimation

When the audio signal spectrum is analyzed with sufficient frequency resolution, formants, individual sinusoids, etc. can be clearly seen, which is referred to as fine-structure spectral envelope in the following. However, with low resolution, it is not possible to observe details, which is referred to as the coarse structured spectral envelope in the following. The noise floor level, although it is not necessarily noise in the definition, throughout the use of the present invention, refers to the ratio of the coarse-structured spectral envelope in the high-resolution spectrum interpolated along the local minimum point to the coarse-structured spectral envelope in the high-resolution spectrum interpolated along the local maximum point. This measurement is obtained by computing a high resolution FFT of the signal segment and applying the peak and valley traces, as shown in fig. 1. Then, a noise floor level is calculated as a difference between the peak trace and the valley trace. This signal is suitably smoothed in time and frequency to obtain a measure of the noise floor level. The peak trace function and the valley trace function may be described in terms of equations 1 and 2,

equation 1

Equation 2

Where T is the delay factor and X (k) is the logarithmic absolute value of the spectrum at the k-line. A pair of two different FFT sizes is calculated, one high and one medium, in order to obtain a good estimate during vibrato and quasi-stationary sounds. The peak and valley traces added to the high resolution FFT are LP filtered in order to discard extreme values. After obtaining two noise floor level estimates, a maximum is selected. In one embodiment of the invention, the noise floor level values are transformed to a plurality of frequency bands, however, other transformations may be utilized, such as curve fitting polynomials or LPC coefficients. It should be noted that several different methods may also be utilized when determining the noise content in an audio signal. However, as mentioned above, the object of the present invention is to estimate the difference between local minima and local maxima in a high resolution spectrum, although this is not necessarily an accurate measure of the true noise level. Other possible methods are linear prediction, autocorrelation, etc., which are commonly used in hard decision Noise/Noise-free algorithms [ "Improving Audio codes by Noise localization," d.schultz, JAES, vol.44, No.7/8, 1996 ]. Although these methods attempt to measure the amount of true noise in the signal, they can be applied to measure the noise floor level defined in the present invention, although they do not give the same good results as the above methods. Analysis by a synthesis method can also be used, i.e. the decoder is placed in the encoder and in this way evaluates the exact value of the amount of adaptive noise required.

Adaptive noise floor addition

In order to add an adaptive noise floor, there must be a representation of the spectral envelope of the signal. This may be a linear PCM value or LPC representation of the filter bank arrangement. The noise floor is shaped according to this envelope before adjusting it to the exact level of the decoder received value. The level can also be adjusted with an additional offset given in the decoder.

In one decoder embodiment of the present invention, the received noise floor level is compared to an upper limit value given in the decoder, and then transformed to several filter bank channels, followed by smoothing in time and frequency by LP filtering, as shown in fig. 2. After the noise floor is added to the signal, the replicated high-band signal is adjusted in order to obtain the correct overall signal level. The energy of the adjustment factor and the noise floor is calculated according to the following equations 3 and 4.

Equation 3

Equation 4

Where k indicates the frequency line, l is the time index of each subband sample, sfb _ nrg (k, l) is the envelope representation, and nf (k, l) is the noise floor level. When noise is generated using energy noiseLevel (k, l) and the high band amplitude is adjusted using adjust factor (k, l), the increased noise floor and high band energy is according to sfb _ nrg (k, l). Fig. 3 to 5 show an example of the algorithm. Fig. 3 shows the spectrum of an original signal that contains a very pronounced formant structure in the low frequency band, and weak formants in the high frequency band. Figure 4 shows the results obtained by processing this signal with SBR without adaptive noise floor addition. It is clear that although the formant structure of the replica high frequency band is correct, the noise floor level is too low. Estimating and adding the noise floor level according to the invention results in the result in fig. 5, where the noise floor superimposed on the replica high frequency band is shown. The advantages of adaptive noise floor addition are both visually and audibly apparent.

Displacer gain adaptation

With multiple substitution factors, the ideal replication process produces a large number of harmonic components, giving a harmonic density similar to the original signal. One method of selecting suitable amplification factors for the different harmonics is described below. We assume that the input signal is a harmonic progression:

equation 5

The substitution of factor 2 results in:

equation 6

It is clear that every second harmonic in the transposed signal has been lost. To increase the harmonic density, harmonics of a higher order permutation (M ═ 3, 5, etc.) are added to the high frequency band. In order to favor most of the higher order harmonics, it is important to adjust their levels appropriately to avoid having one harmonic in the overlapping frequency range dominate. This creates a problem of how to deal with the signal level differences between the ranges of the individual harmonic sources. These signal level differences also tend to vary with program material, making it difficult to use constant gain factors for different harmonics. A harmonic level adjustment method is described herein, which takes into account the spectral distribution in the low frequency band. The outputs from the permuters are fed through gain adjusters and after summing sent to an envelope adjusting filter bank. The low band signal is also sent to this filter bank, which enables spectral analysis. In the present invention, the signal power of the source range corresponding to different substitution factors is evaluated and the gain of the various harmonics is adjusted accordingly. A more elaborate solution is to estimate the slope of the low band spectrum and to apply compensation before input to the filter bank by means of a simple filter arrangement, e.g. a slope filter. It is important to note that this process does not affect the equalization function of the filter bank and the low frequency band of the filter bank analysis is no longer resynthesized from it.

Noise substitution limiting

According to the above equations 5 and 6, the copied high frequency band sometimes contains holes in the frequency spectrum. The envelope adjustment algorithm seeks to make the spectral envelope in the regenerated high frequency band similar to the spectral envelope of the original signal. We assume that the original signal has high energy in one frequency band and that the transposed signal shows spectral holes in this frequency band. Noise or other unwanted signal components can be adjusted to the same energy as the original signal, provided the amplification factor allows taking arbitrary values, which means that very high amplification factors can be added to this band. This is referred to as a substitution of unwanted noise. Order to

P₁＝[p₁₁，…，p_1N]Equation 7

Is the scale factor of the original signal at a given time, and

P₂＝[p₂₁，…，p_2N]equation 8

Are the corresponding scale factors of the permuted signal, where each element in the two vectors represents the subband energy normalized in time and frequency. We obtain the amplification factor required for the spectral envelope adjustment filter bank as follows

Equation 9

By observing G, it is not important to replace certain frequency bands with unwanted noise, since these frequency bands exhibit very high amplification factors relative to other frequency bands. Applying limiters to the amplification factors, i.e. allowing them to change freely to a certain limit value g_maxTherefore, substitution of the unwanted noise can be easily avoided. The use of a noise limiter results in an amplification factor,

G_lim＝[min(g₁，g_max)， …，min(g_N，g_max)]equation 10

However, this expression only shows the basic principle of a noise limiter. Since the spectral envelopes of the transposed and original signals may differ significantly in level and slope, a constant g is used_maxValues are not feasible. Instead, an average gain defined below is calculated

Equation 11

And allows the amplification factor to exceed this value by some amount. To account for wide band level variations, two vectors P may also be used₁And P₂Are divided into different sub-vectors, anTreatments were given accordingly. In this way a very efficient noise limiter is obtained without disturbing or limiting the level adjustment function of the subband signals containing useful information.

Interpolation

In generating the scale factors, the individual channels of the analysis filter bank are typically combined in a sub-band audio encoder. The scale factor represents an estimate of the spectral density within the frequency band that contains the individual channels of the combined analysis filter bank. To obtain the lowest bit rate possible, it is necessary to minimize the number of scale factors transmitted, which means that as large a set of filter channels as possible is used. Typically, this is done by combining the individual frequency bands according to Bark scale, thus employing the logarithmic frequency resolution of the human auditory system. This is possible in an SBR decoder envelope adjustment filter bank, the combination for each channel being the same as the combination used during the scale factor calculation in the encoder. However, by interpolating the values from the received scale factors, the adjusted filter bank can still operate on a filter bank channel basis. The simplest interpolation method is to assign the scale factor value to each filter bank channel within the bank used for scale factor calculation. The permuted signal is also analyzed and a scaling factor is calculated for each filter bank channel. These scale factors and interpolated values representing the original spectral envelope are used to calculate the amplification factors as described above. There are two main advantages to using this frequency domain interpolation method. Compared with the original signal, the replaced signal tends to have a sparser frequency spectrum. Therefore, a smoothing operation of the spectrum is advantageous, which is more effective over a narrow band than a wide band. In other words, the envelope adjustment filter bank may better isolate and control the generated harmonics. In addition, the performance of the noise limiter is improved since spectral holes can be better estimated and controlled with higher frequency resolution.

Smoothing operation

After obtaining a suitable amplification factor, it is advantageous to perform a smoothing operation in time and frequency in order to avoid aliasing and ringing phenomena occurring when adjusting the filter bank and ripples in the amplification factor. Fig. 6 shows the multiplication by an amplification factor with corresponding subband samples. The figure shows two high resolution blocks followed by three low resolution blocks and one high resolution block. It also represents a reduced frequency resolution at higher frequencies. By filtering the amplification factor in time and frequency, for example, using a weighted moving average, the sharp changes in fig. 6 are absent in fig. 7. It is important, however, to maintain the transient structure of the short-time blocks so as not to reduce the transient response of the reproduction frequency range. Similarly, it is important not to filter the amplification factor of the high resolution block excessively in order to preserve the formant structure in the replica frequency range. In fig. 9b, the filtering operation is exaggerated for better visual effect.

Practical embodiment

The invention may be implemented in hardware chips and DSPs in various types of systems for storing or transmitting analog or digital signals, using any codec. Figures 8 and 9 show possible embodiments of the invention. Here, the high frequency reconstruction is done by means of spectral band replication SBR. Fig. 8 shows the encoder side. The analog input signal is fed to an a/D converter 801 and an optional audio encoder 802, as well as a noise floor level estimation unit 803 and an envelope extraction unit 804. The encoded information is multiplexed into a serial bit stream 805 for transmission or storage. Figure 9 shows a typical decoder implementation. The serial bit stream is demultiplexed 901 and the envelope data is decoded 902, i.e. the spectral envelope of the high frequency band and the noise floor level. The decomposed source coded signal is decoded with an arbitrary audio decoder, 903, and upsampled, 904. In this embodiment, an SBR substitution is applied in unit 905. In this unit, the different harmonics are amplified according to the invention using feedback information from the analysis filter bank 908. The noise floor level data is sent to an adaptive noise floor addition unit 906 where a noise floor is generated. According to the invention, the spectral envelope data is interpolated 907, the amplification factor is limited 909, and subjected to a smoothing operation 910. The reconstructed high frequency band 911 is adjusted and adaptive noise is added. Finally, the signal is resynthesized 912 and added to the delayed low band 913. The digital output is converted to an analog waveform 914.

Claims (16)

1. A method for enhancing a source coding system that generates an encoded signal by encoding an original signal, the original signal having a low-band portion and a high-band portion, the encoded signal including the low-band portion of the original signal and not including the high-band portion of the original signal, the method comprising the steps of:

estimating a noise floor level of a highband portion of the original signal, the noise floor level being a measure of a difference between a first spectral envelope determined from each local minimum in a spectral representation of the original signal and a second spectral envelope determined from each local maximum in the spectral representation of the original signal; and

the encoded signals including the noise floor level in the low-band portion and the high-band portion of the original signal are multiplexed to obtain the output signal of the encoder.

2. The method of claim 1, wherein the step of estimating comprises the steps of:

the noise floor level is mapped to a number of frequency bands, and a noise floor level is obtained for each of the number of frequency bands.

3. The method according to claim 1, wherein the noise floor level is represented by linear predictive coding or any other polynomial expression.

4. The method of claim 1, wherein the step of estimating comprises the steps of:

with a sufficient resolution, a fine-structured spectral representation of the original signal is given, in which the harmonic peaks or individual sinusoids are made apparent, the fine-structured spectral representation having local minima and local maxima;

applying a valley trace effect on the spectral representation of the fine structure to interpolate along local minima to obtain a first spectral envelope;

applying a peak trace effect on the spectral representation of the fine structure to interpolate along local maxima to obtain a second spectral envelope;

forming a difference between the first spectral envelope and the second spectral envelope, obtaining a measure of the difference; and

the difference measurement is smoothed to obtain a noise floor level value.

5. A method according to claim 2, wherein the measure of difference is also smoothed over time.

6. A method according to claim 2, wherein the estimating step comprises the steps of:

giving a spectral representation of another fine structure of the original signal using a lower resolution than the resolution used in the giving a spectral representation of the fine structure step;

performing the steps of applying a valley trace effect, applying a peak trace effect, and forming a difference between the first spectral envelope and the second spectral envelope to obtain a further measure of difference; and

selecting between the further difference measure and the noise floor level value, the largest noise floor level estimate is obtained.

7. A method according to claim 1, wherein the spectral envelope of the high-band portion of the original signal is estimated and additionally multiplexed to form the output signal of the encoder for use by a decoding system using high-frequency reconstruction techniques.

8. An apparatus for enhancing a source encoder that generates an encoded signal by encoding an original signal, the original signal having a low-band portion and a high-band portion, the encoded signal including the low-band portion of the original signal and not including the high-band portion of the original signal, the apparatus comprising:

an estimator for estimating a noise floor level of the original signal, the noise floor level being a measure of a difference between a first spectral envelope determined by local maxima in a spectral representation of the original signal and a second spectral envelope determined by local maxima in the spectral representation of the original signal; and

a multiplexer for multiplexing the encoded signals including the noise floor levels of the low-band portion and the high-band portion of the original signal.

9. An apparatus for enhancing a source decoder, which generates a decoded signal by decoding an encoded signal obtained by source encoding an original signal, the original signal having a low-band portion and a high-band portion, the encoded signal including the low-band portion of the original signal and not including the high-band portion of the original signal, wherein the decoded signal is used for high-frequency reconstruction to obtain a high-frequency reconstructed signal including the high-band portion reconstructed from the original signal, the apparatus comprising:

a demultiplexer for demultiplexing an input signal comprising the encoded signal and a noise floor level of a highband portion of the original signal, the noise floor level being a measure of a difference between a first spectral envelope determined by each local minimum in a spectral representation of the original signal and a second spectral envelope determined by each local maximum in the spectral representation of the original signal;

means for obtaining a representation of a spectral envelope of a highband portion of an original signal;

a shaper, which shapes the spectrum of the random noise signal according to the spectral envelope representation of the high-band part of the original signal to obtain a spectrally shaped random noise signal;

a regulator for regulating the spectrally shaped random noise signal according to the noise floor level to obtain a regulated spectrally shaped random noise signal; and

and the adder adds the shaped random noise signal on the adjusted frequency spectrum to the high-frequency reconstruction signal to obtain an enhanced high-frequency reconstruction signal.

10. The apparatus of claim 9, further comprising:

a combiner for combining the enhanced high frequency reconstructed signal and the decoded signal to produce an output signal having a low frequency portion of the original signal and a high frequency portion of the original signal reconstruction.

11. The apparatus of claim 9, further comprising:

a high frequency reconstruction module for generating a signal, the high frequency reconstruction module having a summer for summing a number of high frequency reconstructed signals from different low band frequency ranges from the decoded signal to obtain a signal to be generated; and

an analyzer for analyzing a low-band portion of the decoded signal and for providing control data to the summer.

12. A method for enhancing a source decoding system that generates a decoded signal by decoding an encoded signal obtained by source encoding an original signal, the original signal having a low-band portion and a high-band portion, the encoded signal including the low-band portion of the original signal and not including the high-band portion of the original signal, wherein the decoded signal is used for high-frequency reconstruction to obtain a high-frequency reconstructed signal including the high-band portion reconstructed from the original signal, the method comprising the steps of:

demultiplexing an input signal comprising the encoded signal and a noise floor level of a highband portion of the original signal, the noise floor level being a measure of a difference between a first spectral envelope determined by each local minimum in a spectral representation of the original signal and a second spectral envelope determined by each local maximum in the spectral representation of the original signal;

obtaining a spectral envelope representation of a high-band portion of an original signal;

shaping the spectrum of the random noise signal according to the spectral envelope representation of the high-band part of the original signal to obtain a spectrally shaped random noise signal;

adjusting the spectrally shaped random noise signal according to the noise floor level to obtain an adjusted spectrally shaped random noise signal; and

adding the adjusted spectrally shaped random noise signal to the high frequency reconstructed signal to obtain an enhanced high frequency reconstructed signal.

13. A method according to claim 12, wherein the spectral envelope representation comprises an energy measure of the energy of the high frequency reconstructed signal and the noise floor, the method further comprising the steps of:

the high frequency reconstructed signal is adapted such that the combined energy of the high frequency reconstructed signal and the adapted spectrally shaped random noise signal equals the energy measure represented by the spectral envelope.

14. A method according to claim 12, wherein the step of adjusting the spectrally shaped random noise signal comprises the step of smoothing the spectrally shaped random noise signal in time and/or frequency.

15. A method according to claim 12, wherein the spectral envelope of the high frequency reconstructed signal is adjusted by interpolation.

16. A method according to claim 12, wherein the spectral envelope of the high frequency reconstructed signal is adjusted by smoothing with an envelope adjustment amplification factor.