JP4468580B2 - Method and apparatus for level compensation for input signals - Google Patents

Description

【００５４】
５ パラメータの各増分で望ましい平均相対ゲインを決定する。等式Ｅｑ．１を使用して、表１の各縦列の平均が計算されうる。これは各周波数のゲインの範囲を計算するためにも有用である。等式Ｅｑ．２はまず相対ゲインデータを調整して合計平均点の周囲に集め、偏差を最適化する。等式Ｅｑ．３及びＥｑ．４は調整された縦列の最大及び最小値を計算する。その結果は図５に示される。
【００５５】
【数１】

【００５６】
６ この例のアナログ回路では、周波数設定はポテンショメータによるゲイン設定に相関しなければならない。例えば、デュアルポテンショメータを使用すると、ゲイン回路（図４）を追加することができるので、ゲインと周波数が単一のつまみから同時に制御できる。等式Ｅｑ．５及びＥｑ．６は、この例ではノッチ周波数、ゲイン、及び抵抗（ｆ（ｘ）、Ｇ（ｘ）、及びＲ（ｘ））の間の関係を記述する。結果は表２に示される。
【００５７】
【数２】

【００５８】
【表２】

【００５９】
公式６では、抵抗器Ｒ４の値は図５で示される曲線の形状を決定する。
７ ゲイン設定と周波数設定の最も近いマッチ（match)を見い出す。周波数回路は機能的に最上の配置に設定として予め決定されているので、ゲイン回路を調整してそれにマッチさせなければならない。抵抗値Ｒ４を適切に選択することによって、図５の線に見られるような結果として得られるゲインの変更は、図５に菱形によって示されるような周波数に対するゲインの望ましい変更に近似することができる。抵抗Ｒ２は全体のゲインを設定する。
８ 他のパラメータについて繰り返す。
【００６０】
ステップ６での例示方法の代替案には、フィルタ回路（図５Ａ、また図６及び図７参照）及びデジタル版（図８参照）を操作することによる周波数変更を伴う統合ゲイン変更が含まれる。デジタル版の１つの変形はメモリ中に格納された平均ゲインデータを有し、測定点の間の周波数でゲインを補間する。もう１つの変形はｆ（ｘ）とＧ（ｘ）の間の関係を使用してゲインを計算する。
【００６１】
図５Ａは相互連係の深さ（デプス）及び周波数レベル補償を伴うノッチ・フィルタを示す。オーディオ周波数に対するゲイン伝達関数(gain transfer function)は等式Ｅｑ．７によって示される。
【００６２】
【数３】

【００６３】
フィルタは、ポテンショメータの単一の要素が周波数とゲインの両方を変更できるように配置される。図６は標準フィルタ４１、すなわちサレン・キー２次ハイパスフィルタ(Sallen-Key second-oder high pass filter)の機能図を示すが、これは、図７の要素４１でＲ２１及びＲ２２によって示されるデュアルポテンショメータの周波数−ゲイントラッキング(frequency-gain tracking)を生じるよう操作されている。普通、出力＃２（図６、要素４３参照）はグランドに接続される。その代わりに、Ｒ２２を図７の電流増幅器ＡＲ６に接続すると、要素４４では出力＃１（図６、要素４３参照）で同じ伝達関数と、電流がＲ３＋Ｒ２２に反比例する第２出力が可能になる。Ｒ＝Ｒ４＋Ｒ２１＝Ｒ３＋Ｒ２２及びＣ＝Ｃ１＝Ｃ２のとき、２つの出力の伝達関数はどちらもｆ＝１／（２πＲＣ）及びＱ＝０．７０７の極でハイパスすることに注目するべきである。ミクサ(mixer)の高周波数出力ゲインは−Ｒ７／（Ｒ８＋Ｒ）である。従って、ハイパスフィルタの周波数とゲインは前記ポテンショメータの単一の制御要素によって制御されうる。図８を参照すると、レベル補償のデジタル版が周知のシステムに次いで示される。当業技術では、関数を選択する（例えば、ハイパスフィルタを設定する）ユーザ・インタフェースを提供し、周波数に関する入力（例えば、ハイパスフィルタのカットオフ周波数）を提供することは周知である。望ましい関数を実現するために必要なフィルタ係数は入力周波数に基づいてメモリから選択される。その出力はフィルタ周波数の結果として得られる調整である。本発明の実施形態によれば、メモリはさらに、ユーザからの周波数入力に基づいて選択されるゲイン値を格納するために使用される。従って、レベル補償は、上記で詳細に説明されたように、選択された周波数に基づいて行われる。
【００６４】
レベルの変化を発生する各制御を別個に補償する必要の他に、この解決法では十分でない状況が存在する。（音源とイコライズ・タイプの組み合わせが一貫している）予想できる状況では、上記の処理は非常に良好な結果を生じる。音源材料の種類が制限されているほど、またイコライズ要素の範囲／効果が制限されているほど、経験的に導出される補償されたプロセッサを開発することは容易になる。しかし、音源が多様になるほど、必要な試験は多くなり、結局、試験結果が矛盾するようになる。
【００６５】
最も複雑な状況は、（フレッチャー・マンソン曲線は平均値であるので各人はそれから見て異なっていることを認めた上で）全てのありうる人間の聴取者に対し、全てのありうる音源（話(speech)、音楽、効果音(sound effect)等）に対して、広い範囲を（例えば、２０〜２０ｋＨｚの範囲の周波数制御）を有するつまみを伴う設計を行うことであろう。これは音源の変化が所与のイコライザと干渉するため実際には不可能である。例えば、オーケストラのような幅広い音源に対して良好に動作する低域強調周波数つまみ(bass boost frequency knob)を設計すると想定し、低域で、補償回路増幅レベルが増大する（これはフレッチャー・マンソン効果のため行われる見込みが大きい）と想定する。音楽のトライアングルだけが演奏される音楽の箇所があることが想定される。低域でイコライザつまみを調整する場合、トライアングルは強調されうる低域周波数を有さないので、トライアングルのトーンに対する影響はない。変化は全く聞き取れないと予想されるが、ゲイン補償器はレベルを追加し続けるので、つまみを前後に回すとトライアングルの音量は変化する。トライアングル（または低域周波数を有さない任意の音声）の場合、このつまみは音量調整に変わり、ユーザを混乱させる。すなわち、（上記のものを含む、任意の方法によって決定されるような）イコライズつまみの設定位置だけに基づいて補償を行う回路は、大幅に異なった状況では良好に動作しないであろう。こうした状況では、上記で説明された第１実施形態が好適である。
【００６６】
本発明の第３実施形態が、ダイナミック処理補償回路として図１３〜図１６に示される。ダイナミック・プロセッサは、ピークを制限して過負荷を防止する、信号のダイナミックレンジを圧縮する、レンジを拡大する、音(note)を長く持続する、等の任意の数の効果を達成するため、信号のゲインを変更する。合計ゲインは入力信号レベルに基づいて変わるので、上記で説明された音源調整補償方法(source-adjusted compensation method)を使用してパラメータの調整を補償するのは非常に困難である。このような方法はプロセッサの作業を取り消しがちである。その代わり、本発明の実施形態によれば、予想されるゲインの変更を以下の手順で補償することができる。
１．処理の後望ましい正規化された聴取レベルを想定する。最も簡単な正規点は０ｄＢＶである。
２．ゲイン１で圧縮しない場合、０ｄＢＶの入力信号は０ｄＢＶの出力を生じるので、０ｄＢＶ入力信号の圧縮からのゲインの変化は、０ｄＢＶ出力を維持するためには等しくて反対のゲインの変更によってマッチされるべきである。例えば、９０ｄＢから６０ｄＢへのダイナミックレンジの低減が望ましい場合、−９０ｄＢＶから始まって、入力信号が０ｄＢＶまで上昇するにつれて、徐々にゲインを低下させるよう圧縮器が使用される。しかし０ｄＢＶ入力の出力は−３０ｄＢＶとなるであろう。−９０〜−３０ｄＢＶでなく、−６０〜０ｄＢＶの範囲が望ましいことの方が見込みが大きい。従って、ゲインは３０ｄＢだけ増大されるべきである。
３．ゲイン調整は図１３に示される。正及び負の制御電圧入力を有する電圧制御増幅器（ＶＣＡ）を使用して、０ｄＢＶの入力を有する他の端子で予想される制御電圧に等しい一定の制御電圧が１つの端子に供給される。例えば、あるパラメータ設定での０ｄＢＶ入力信号は負制御ポートで＋１００ｍＶを発生して、ゲインを低減する。正制御ポートでの＋１００ｍＶ電圧はこの低減を打ち消す。また、等しい反対の制御電圧が同じ回路の単一の端子で結合されることもある。
【００６７】
この方法を実現するため、ＲＭＳ出力信号と、正規化された入力レベル（この場合０ｄＢＶ）でのＲＭＳ出力に等しい定電圧源との両方を修正するために同一の回路が使用される。この実施形態では、しきい値及び比のつまみに対してデュアルポテンショメータが使用され、両方の制御電圧回路を同時に調整することができるので、２つの制御電圧はタンデムに変化する。同じ機能を達成するため他の回路が使用されることもできる。
【００６８】
図１４を参照すると、通常の圧縮器の伝達関数Ｖ_in対Ｖ_outのグラフが示される。低いしきい値と大きな比の場合どれだけピーク・レベルが低下するかに注意されたい。図１５では、レベルを０ｄＢＶに正規化した様々な設定でのレベル補償された圧縮器のグラフが示されている（図１５Ａでは、１４ｄＢＶに正規化された同じ効果が示される）。例えば、２：１の比では、−１６ｄＢＶしきい値の圧縮は０ｄＢＶ入力でゲインを−８ｄＢ低下させる。従ってゲインは＋８ｄＢ補償される。∞：１では、ゲインは＋１６ｄＢ補償されなければならない。ダイナミックレンジの高いレベルで正規化すると、一般に０ｄＢＶ以下にある信号の大部分の音量を増大させることになることに注意すべきである。しかし、目的が単に「上を刈り込む」ことであるのでないならば、このゲイン調整は普通望ましい。図１６では、ＴＨＡＴ４３０１ダイナミック・プロセッサ・チップを使用するダイナミック・プロセッサの詳細な回路図が、ＲＭＳ変換器及び対数ベース(log-based)のＶＣＡと統合して示される。
【００６９】
本発明の実施形態はシグナルプロセッサ中の変化によって発生したレベルの差を補償する。補償器はプロセッサの信号の修正に基づいてレベルを修正するのみであるべきである。レベル補償の誤った起動を回避するため、検出回路またはスイッチが、シグナルプロセッサが調整されていないときは、レベル補償器をその修正の変更から凍結するべきである。
【図面の簡単な説明】
【図１】 本発明の実施形態によって構成される装置の全体構成図である。
【図２】 本発明の第２実施形態の全体構成図である。
【図３】 図２のプロセッサの詳細な回路図である。
【図４】 図２の増幅器の詳細な回路図である。
【図５】 図２〜図４の動作の例における周波数とレベル補償の間の関係を示すグラフである。
【図５Ａ】 本発明の実施形態によって構成される相互連係の深さ（デプス）及び周波数レベル補償を伴うノッチ・フィルタである。
【図６】 本発明の実施形態によって構成される、ポテンショメータの単一要素が周波数とゲインの両方を変更できるレベル補償装置の構成図である。
【図７】 図６の装置の詳細な回路図である。
【図８】 本発明の実施形態によって構成されるデジタル・レベル補償システムの構成図である。
【図９】 本発明の第１実施形態によって構成される装置の構成図である。
【図１０】 図９の装置の詳細な例である。
【図１１】 図１０のいくつかの要素のさらに詳細な回路図である。
【図１２】 図１１のいくつかの要素のさらに詳細な回路図である。
【図１３】 本発明の第３実施形態の構成図である。
【図１４】 当業技術で周知の圧縮の効果を示すグラフである。
【図１５】 本発明の実施形態によるレベル補償を伴う圧縮の効果を示すグラフである。
【図１５Ａ】 異なったレベルに正規化された図１５と同様のグラフである。
【図１６】 本発明の第３実施形態を実現するダイナミック・プロセッサの詳細な回路図である。[0001]
(Background of the Invention)
The present invention relates to a method and apparatus for performing level compensation on an input signal. In particular, the present invention relates to a method and apparatus for adjusting the output level of an input signal to compensate for level changes made to the input signal by a signal processor.
[0002]
A category of devices called “signal processors” are well known in the art. Such a processor can be used to modify input signals such as music (or other audio speech) converted to electronic analog or digital signals. Typical audio processors include timbre correction (eg, equalizer or EQ, vocoder, distortion effect, chorus effect, flanger effect, ring modulator, wah-wah effect), dynamic correction (Compression, expansion, tremolo, vibrato, etc.), reverberation, delay, many other modifications, and combinations thereof. Other processors are well known in the art, such as general video processors (eg, color filters) and other processors used in various fields. All of these processors can have the potential “side effect” of changing the absolute peak level of the original input signal with respect to other signals or devices present. This change in peak level is usually undesirable and is often not readily apparent to the operator, and can often easily affect the operator's judgment without the knowledge of the operator.
[0003]
In the audio field, there are three common reasons to “equalize” speech: 1) remove problematic noise (eg remove unpleasant air conditioner hum), 2) Create a pleasant tone (eg, adding a low range to make the singer's voice a pleasant “husky” voice), and 3) create an effect (eg, the Beatles “Yellow” As in the case of “Yellow Submarine”).
[0004]
When using an equalizer to enhance or block part of the sound for some purpose, the volume of the equalized sound is actually increased (if emphasized) or decreased ( When blocking). As is well known in the speech engineering art, it is important to allow comparison between the original and equalized speech as part of the process of using an equalizer (or other speech processor) . This is accomplished by a “bypass” switch that selects whether the audio goes through the equalizer circuit or bypasses it.
[0005]
Acoustic psychology concerns and results : Research on how people hear sounds is generally called psychoacoustics. Phenomena well known in the art are first organized by Fletcher and Munson and most commonly displayed as a graph called the “Fletcher Manson curve” or equal loudness contour. This data shows that the human ear and brain organs interpret the loudness as a function of frequency, with the following two specific effects.
[0006]
1—Human hears sound at different frequencies non-uniformly (at the same volume level, low and high frequency sounds seem quieter than mid frequencies), and
2- As the overall sound becomes louder, it becomes gradually more uniform. In the case of very loud speech, the above effect is reduced or disappears. In other words, when the volume is increased, people can hear more bass and treble, which is why some people listen to music with very loud sounds.
[0007]
Humans can usually hear audio within the frequency range formed by the human voice (approximately 100 Hz to 4 kHz) more easily than higher or lower frequency audio. Both low bass guitar notes and high violin sounds must be played with much greater energy than the voice if you want them to be as loud as the voice.
[0008]
The unique effects outlined above interact with each other. As the sound gets louder, the frequency difference does not affect listening, and at very large levels, the amount of energy for bass, voice and violin that sound the same magnitude as each other is about the same.
[0009]
This is generally experienced when using stereo equipment. When the volume is at a normal listening level, the compact disc has acceptable sound. When the volume is lowered too much, the music suddenly sounds like there isn't enough bass and not enough cymbals or other high frequencies (the Fletcher Manson curve shows how small the low and high frequencies are. Describes). Stereo equipment is often supplied with “loudness” control (on / off switch or variable knob) to compensate for this (very limited success) But only.) Since the Fletcher Manson curve is the average of data empirically derived from many people, it is an approximation for the experience of any given individual.
[0010]
Audio engineers have worked on this phenomenon many times. Below are two examples of how this problem is addressed in the art.
[0011]
The first example relates to an instrument that produces only one sound called tom-tom on a drum set. When recording a tam tom, the audio engineer uses an equalizer to try to make it sound better. For example, a pleasant voice can be heard near the fundamental resonance of the drum between 400 and 800 Hz. There is also a very unpleasant voice just above this frequency range. The engineer can use a low pass equalizer (which cuts off the high frequency signal) with control to select the frequency (which can select a point to cut off the sound above it). This allows the engineer to cut off the uncomfortable range and then fine tune the low frequency to find the spot where the drums are best heard.
[0012]
Initially, the technician cuts off unpleasant frequencies above the pleasant range. However, since the sound becomes very small, the volume is increased greatly. The engineer then fine tunes to get the best sound. The drum itself has a different degree of loudness at each frequency, and the best position to set the control is in the frequency range where the drum sound is relatively low. If the sound of the best part of the drum sound is very small, it may become unperceived (again due to the Fletcher Manson effect). From experience or training, technicians increase (again) the volume to help them hear the best voice. A bypass switch is then used to see if the equalized signal is improving over the original (unequalized) speech. When the unequalized sound becomes too loud, it becomes impossible to determine what the engineer has achieved (as explained by the Fletcher Manson curve, just because it is louder, (Almost always sounds rich and good), and the sound is too loud and may hurt the engineer's ears (because the equalized speech has to be loud enough to be heard). Further, when the volume is increased, the speaker may be damaged because the sound is too loud.
[0013]
In the second example, the singer's recorded track is mixed with the rest of the band (drum, bass, guitar, piano, etc.). As explained above, music is heard “bassier” when the volume is raised, as explained by the Fletcher Manson curve (or lower volume means “less bass” (sounds like “bass“ bassy ”)”). As more bass is added to the audio by the equalizer, the total volume increases.
[0014]
If the voice recording sounds “thin” due to poor microphone placement, the technician may try to compensate by raising the low range equalizer. The sound of the voice will improve, but the extra low will make the voice louder than the rest of the band, so it will be lowered to an appropriate level. The result is that there is no longer enough bass in the voice (due to the Fletcher Manson effect). Therefore, this process is repeated several times until a satisfactory result is achieved.
[0015]
Known techniques for solving the problems addressed above are described in the following complex procedure.
[0016]
1—Select the channel to be equalized (“EQ Ch.”).
[0017]
2- Set up another channel that duplicates the original unequalized input (“UNeq Ch.”).
[0018]
3-EQ Ch. Listen only (ie, 'SOLO' EQ Ch.-Listen only to that channel and turn off all other channels).
[0019]
4-EQ Ch. Adjust the equalizer settings for until satisfied.
[0020]
5-Solo EQ Ch. And UNeq Ch. Alternately. That is,
a-EQ Ch. Output level of UNeq Ch. Matches the output of.
[0021]
b-EQ Ch. Adjust the equalizer settings for.
[0022]
6-When mixing (with other audio channels turned back on)
a-EQ Ch. And UNeq Ch. Set both levels (at one time).
[0023]
b-EQ Ch. And UNeq Ch. To determine and adjust the equalizer settings.
[0024]
7—Adjust as needed, repeat steps 3-6 as needed.
[0025]
The above solution is not only complex, but many performers and many audio engineers do not understand or recognize this problem or how to compensate for it. This is because the amplified sound of many clubs, concerts, weddings, etc. is often poor (eg harsh, piercing, bass too strong, too loud, etc.) it is obvious.
[0026]
In the art, there are two very common devices (in addition to equalizers, reverberators, and other processors) that actually change the volume of the source matarial. Such devices do not make moment-to-moment changes that affect the signal.
[0027]
1-Loudness Control (a switch or knob found in many home listening devices) is an attempt to compensate for the Fletcher Manson effect at low listening levels. This is a special tone adjustment that emphasizes both low and high frequencies with a single control (but the low and high frequencies emphasized by this are the same as normal low and high frequency controls). Somewhat different).
[0028]
A 2-AGC (Automatic Gain Control) amplifier circuit is used to compress (reduce) the dynamic range of a signal. This means that no matter how loud (or soft) the original signal (input) is, the output is always at the same volume. This is usually done with household products (VCRs and inexpensive cassette tape recorders) with low-quality built-in microphones, and devices that deliberately sacrifice audio quality for call clarity (CB and amateur radio, Part of the phone). The output does not depend on either the sound source or any particular setting. Audio fidelity is deliberately sacrificed to overcome the limitations of noisy environments or to allow the use of inexpensive equipment. In practice, such devices usually do not even have input volume adjustment, but some high quality products can choose between using an AGC circuit or an actual input level control circuit. Such a device has an on / off switch for the AGC circuit. When AGC is turned on, the actual input circuit is bypassed, and when AGC is turned off, the AGC circuit is bypassed.
[0029]
AGC is generally associated with a dynamic processor. This is a professional device that allows the user to control the degree to which dynamic range reduction occurs. With AGC, these devices actually change the dynamic range of the input signal.
[0030]
In view of the above, intelligent level corrections can be made to compensate for level changes caused by the processor, and the use of the processor can be made without the need for compensation or even need to understand the need for compensation. There is a need for a method and apparatus that allows better selection by different users.
[0031]
(Summary of Invention)
These and other needs are met by the method and apparatus of the present invention. In accordance with embodiments of the present invention, an intelligent correction of the level of the output signal is made to compensate for peak level changes.
[0032]
In the first embodiment of the present invention, the unprocessed sound source signal is first compared with the final processed signal to determine the amount of change, and then the level adjustment is performed using the comparison. Standard components are used,
1—Compare the signal before the processor (preprocessor signal) (original unprocessed signal) with the signal after the processor (after processing) (postprocessor signal),
2- Record the level difference between them, then
3—compensate the processed signal, for example by matching the signal level after the processor with the signal level before the processor;
Build a stand-alone device.
[0033]
In this embodiment, one purpose is to bring about a single overall level change, and no matter how many processors are used in combination, the resulting final level change is one, so this The circuit will work for any combination of changes. Thus, only one circuit is required for a signal path, regardless of the number of processors between the “pre-” and “post-” reading points.
[0034]
The processed signal may also be provided to the level compensator, and the output of the level compensator and the unprocessed signal may be provided as inputs to the comparator. The output of the comparator then provides a feedback signal to the level compensator. Therefore, the level compensator modifies the level of the processed signal until the output of the level compensator becomes equal to the level of the unprocessed signal.
[0035]
In the second embodiment of the invention, the compensation provided to the processed signal is derived empirically rather than by direct measurement. The actual level change that occurs as a result of the operator using a given processor control is measured. Each control over the processor that "generates" an undesirable level change is also used to control the circuitry that provides level compensation by producing the desired reverse level change. This embodiment is suitable for situations where the range of changes in the characteristics of the signal to be processed is known and the range of changes in the processor control settings is known. According to this embodiment of the invention, the development of compensation for a single control of the processor includes the following steps.
1. Adjust the control of the processor in a sufficiently small step (for example, 30 degree rotation for a rotary knob) while listening to a typical signal source while listening normally.
2 For each position of control, the resulting change in overall signal level is recorded (by an objective observer and / or an experienced observer to match the perceived level).
3 Repeat with a wide variety of sound source materials.
4 Edit the results and design a circuit that counteracts the level changes caused by processor control changes. This circuit should be adjusted by the same controller that adjusts the processor parameters. Two implementations of using a standard potentiometer as the controller are as follows.
[0036]
a. For example, having a separate gain circuit in series with the processor whose gain is adjusted by a separate potentiometer element coaxial with the processor controller.
b. Modify the processor to incorporate the desired gain change.
The ability to individually adjust the compensation for specific situations and / or uses is a unique advantage of this method. Another advantage is that the circuitry required to achieve is minimal (and therefore inexpensive).
[0037]
In a third embodiment of the present invention, a method is provided that compensates for level changes when the processor itself makes dynamic level changes, such as, for example, a compressor, expander, and the like. While it is desirable for them to change the dynamic range of the sound source material, it is usually undesirable for these devices to change the peak level (the loudest spot) for the rest of the acoustic environment.
[0038]
In audio, such devices typically operate by comparing the level of the input signal to a threshold, which is a constant reference voltage that is set by the user using a controller. . When the signal source crosses a set threshold level, the signal is processed through a single control that sets the ratio of processing that is most often applied (for example). With any of these two control settings, the total effect on the signal peak (maximum instant) is simply the difference from the value multiplied by the threshold and ratio settings in decibels. Function. This function is often expressed as a control voltage and is generally applied to devices such as a VCA (voltage controlled amplifier) that constantly performs the desired change in level.
[0039]
This method embodiment uses the same control settings for the action of the processor operating the second control signal path using a separate constant reference voltage as input. The first control signal is responsive to the audio input signal. The second control voltage depends on the control knob of the processor, but not on the input signal, resulting in the desired constant level of compensation. This compensation control voltage is applied to the same device (eg, VCA) used by the processor. Also, since the amount of level compensation is determined by the same control that sets the change in peak level output, the amount of level compensation automatically tracks (and accurately compensates for) changes in processor settings. The result is an integrated processor and level compensation system.
[0040]
(Detailed explanation)
Referring to FIG. 1, a block diagram of an apparatus for performing level compensation according to an embodiment of the present invention is shown. Level compensation refers to performing one of the following steps.
[0041]
1. As the processor reduces the overall level of the input signal, the apparatus of the present invention increases the gain to match the original signal level.
[0042]
2. As the processor increases the overall level of the input signal, the apparatus of the present invention reduces the gain to match the original signal level.
[0043]
This allows the user to make decisions about the effectiveness of the processor without being “fooled” by signal level differences. This can be done without any user involvement (this is a monotonous and tedious process as explained above). A switch is available that disables the circuit if desired. As will be described in the following embodiments, the present invention can be applied to compensate for the volume of an audio signal. As those skilled in the art will appreciate, the present invention (eg, the circuit of FIG. 1) is intended to compensate for level changes caused by processors of other types of signals (eg, scanning video signals, MRI, CAT). Can be applied.
[0044]
Referring to FIG. 1, an input signal 10 is provided to a processor 11 (eg, a digital signal processor) whose signal is modified. For example, the signal is modified by the processor by reducing the level of a particular frequency band of the signal (eg, as in an audio filter). The original raw signal and the modified signal from the processor (ie, the modified signal) are provided to the comparator 13 (ie, the modified signal is provided either before or after the level compensator 14). . The comparison device 13 outputs a difference signal indicating the difference between the original signal and the corrected signal. The level compensation circuit 14 receives the modified signal and the difference signal, and performs level compensation on the modified signal (for example, corrects an increase or decrease in gain generated by the processor 11). If desired, the level-compensated signal (level-compensated signal) can be provided to an output device 12, such as a recording device, a speaker or the like.
[0045]
As indicated above, an example of a processor is one that changes tone quality like an equalizer. Sound quality, also called timbre, is the result of a balance of different frequencies measured in hertz (cycles / second). Human hearing is limited to a range between 20 Hz and 20 KHz (20,000 Hz), also called the sound spectrum. An equalizer, tone adjustment, filter, and the like are devices that change sound quality. These all operate by allowing the volume of a selected portion of the audible frequency range to be increased or decreased. A device that raises or lowers the entire frequency range uniformly is simply volume control. The equalizer operates as a volume control for a limited part of the listening range.
[0046]
A first embodiment of the present invention is shown in connection with FIGS. Referring to FIG. 9, a block diagram of this embodiment is shown. In FIG. 10, a more detailed example of this circuit (ie, a Root-Mean-Square level compensation system) is shown. FIG. 9 shows a post processed signal (“audio signal”) from the signal processor 61, a pre-equalized “control” signal 10, and a post input from the RMS level compensator 62. Three signals, the compensator “control” signal, are input to the RMS level compensator 62. The RMS levels of the input signal and the control signal are respectively measured by the RMS converter 72A and the RMS converter 72B (FIG. 10) and compared (for example, by the window comparator 73). If the control signal is greater than the output signal, then the signal processor 61 reduces the signal level at that time, so the gain of the comparator is increased by adjusting the digital control resistor 78. If the control signal is small, the gain is reduced. The rate of change of gain is determined by the frequency of the oscillator 75 and the resolution of the resistor. The gain changes until the output level matches the control signal level or until the gain limit is reached. If there is no output signal detected upon sensing by the signal detector 76 and the mode of operation is a “hold” mode (as determined by a Match / Hold Switch 77), The resistor value does not change and the gain remains the same.
[0047]
Referring to FIG. 11, a more detailed schematic of the circuit of FIG. 10 is shown, where the block shown as DS1666 operates as a digitally controlled attenuator and is limited in scope by resistor R2. The audio signal can be boosted by the output gain stage of the operational amplifier AR2. The control inputs to the DS 1666, up / down, increment, and chip selection are operated by “window comparator”, “oscillator” and “disable” circuits, respectively. The “RMS converter (RMS-conv)” converts the “control (CTRL)” signal and the “output (OUT)” signal into an RMS log-equivalent signal. A “hysterisis” circuit determines whether “CTRL” is sufficiently larger or smaller than “OUT”. If there is a large difference, the “oscil” increments or decrements DS1666. Its direction is determined by the upper “hysteresis” circuit output. XOR gate U3 provides the necessary timing of the chip select input along with the incremental input.
[0048]
Referring to FIG. 12, a circuit diagram of the subcircuit of FIG. 11 is shown. RMS conversion is achieved by THAT2252 log converter and rectification (via diode D1 and capacitor C3). Resistor R16 provides its own release time by discharging capacitor C3. The oscillator is a standard 555 timer circuit activated by a reset pin. Note that the comparator VR1 in the hysteresis block has a degree of hysteresis that allows an approximate level match, for example within ± 2 dB. The signal detection VR2 is responsive to a negative going AC signal that falls below a preset noise threshold. This serves as a safety feature that prevents the compensation circuit from accidentally responding to improper signals, especially during the recovery time of the RMS converter.
[0049]
A possibly more accurate version of this embodiment of the present invention uses the same steps as above, but the comparison includes analysis of the frequency spectrum of the signal before and after processing. The applied level compensation is based on the signal level difference and the frequency at which the level difference occurs. For example, Fast Fourier Transform (FFT) is used, but other frequency band inspection techniques are possible. An example of this procedure is shown below.
1-FFT of pre-processing signal and post-processing signal is performed.
2- Both RMS values are derived for the selected bandwidth. In typical audio work, 1/3 octave is common, which would calculate about 30 bands for each of the two FFTs. Even less bandwidth may be sufficient, especially if determined carefully.
3 “Weigh” the result of step 2 with an “average” Fletcher-Munson curve.
4- Sum up the results of step 3 to give a subtotal before and after processing.
5- Take the difference between the subtotals in step 4 (subtract the other from one).
6 Adjust the final output level according to the result of step 5.
[0050]
The purpose of this embodiment of the present invention is a single overall level change, and no matter how many processors are in the combination, the result is only one final level change. Works for any combination of changes. Thus, only one circuit is required for a given signal path, regardless of the number of processors between “previous” and “after” reading points.
[0051]
A second embodiment of the present invention is shown in FIGS. 2-8, where level compensation is derived empirically. Derivation of integrated level compensation for the audio signal processor depends on the characteristics of the signal to be processed and the estimation of the level to be compensated. That is, changes in processor settings must be assumed to affect that level as well, since all its expected input signals are similar. For example, a notch filter can be used to modify the cymbal sound, and setting the notch frequency to 400 Hz will increase the hearing level by 6 dB over the 1.25 kHz setting for most cymbals. Become. To maintain equal levels, the gain must be compensated by -6 dB when changing the 400 Hz frequency setting to 1.25 kHz. Referring to FIG. 2, an overview of how frequency and gain are tracked via a dual potentiometer 24 is shown.
[0052]
An example of the method of this embodiment is shown below.
1 Select the desired processor. In this example, it is a notch filter (as shown in FIG. 3) having a variable frequency (400 to 1250 Hz) and a fixed depth (depth).
2 For example, select a typical sound source when hitting a cymbal. For example, a starting parameter and an output level having a notch frequency of 1.25 kHz and an output gain reference of 0 dB are set.
3. Change the parameter by the selected increment and adjust the gain to match the previous level if necessary. New parameter settings and new gain settings are recorded. Iterate over the useful range of variable parameters. For example, as shown in Table 1, row 1 shows the relative gain change required to maintain an equal level of cymbal speech at 1/3 octave intervals from 400 Hz to 1.25 kHz.
4 Repeat steps 2 and 3 for many similar sound sources. Table 1 is data from 10 similar cymbals.
[0053]
[Table 1]

[0054]
5 Determine the desired average relative gain at each parameter increment. Eq. 1 can be used to calculate the average of each column in Table 1. This is also useful for calculating the gain range for each frequency. Eq. In step 2, the relative gain data is first adjusted and collected around the total average point to optimize the deviation. Eq. 3 and Eq. 4 calculates the maximum and minimum values of the adjusted column. The result is shown in FIG.
[0055]
[Expression 1]

[0056]
6 In this example analog circuit, the frequency setting must correlate with the gain setting by the potentiometer. For example, if a dual potentiometer is used, a gain circuit (FIG. 4) can be added so that gain and frequency can be controlled simultaneously from a single knob. Eq. 5 and Eq. 6 describes the relationship between notch frequency, gain, and resistance (f (x), G (x), and R (x)) in this example. The results are shown in Table 2.
[0057]
[Expression 2]

[0058]
[Table 2]

[0059]
In Formula 6, the value of resistor R4 determines the shape of the curve shown in FIG.
7 Find the closest match between the gain setting and the frequency setting. Since the frequency circuit is predetermined as a setting in the highest functional arrangement, the gain circuit must be adjusted and matched. By appropriately selecting the resistance value R4, the resulting gain change as seen in the line of FIG. 5 can approximate the desired change of gain for frequency as shown by the diamonds in FIG. . The resistor R2 sets the overall gain.
8 Repeat for other parameters.
[0060]
An alternative to the exemplary method at step 6 includes an integrated gain change with a frequency change by manipulating a filter circuit (see FIG. 5A and also FIGS. 6 and 7) and a digital version (see FIG. 8). One variant of the digital version has average gain data stored in memory and interpolates the gain at the frequency between the measurement points. Another variation uses the relationship between f (x) and G (x) to calculate the gain.
[0061]
FIG. 5A shows a notch filter with interworking depth (depth) and frequency level compensation. The gain transfer function with respect to audio frequency is the equation Eq. Indicated by 7.
[0062]
[Equation 3]

[0063]
The filter is arranged so that a single element of the potentiometer can change both frequency and gain. FIG. 6 shows a functional diagram of a standard filter 41, a Sallen-Key second-oder high pass filter, which is a dual potentiometer indicated by R21 and R22 in element 41 of FIG. Is operated to produce frequency-gain tracking. Normally, output # 2 (see FIG. 6, element 43) is connected to ground. Instead, connecting R22 to the current amplifier AR6 of FIG. 7 allows element 44 to have the same transfer function at output # 1 (see element 43, FIG. 6) and a second output whose current is inversely proportional to R3 + R22. It should be noted that when R = R4 + R21 = R3 + R22 and C = C1 = C2, the two output transfer functions are both high-pass at the poles of f = 1 / (2πRC) and Q = 0.707. The high frequency output gain of the mixer is -R7 / (R8 + R). Thus, the frequency and gain of the high pass filter can be controlled by a single control element of the potentiometer. Referring to FIG. 8, a digital version of level compensation is shown next to a known system. It is well known in the art to provide a user interface for selecting a function (eg, setting a high pass filter) and providing an input related to frequency (eg, a cutoff frequency of the high pass filter). The filter coefficients necessary to implement the desired function are selected from memory based on the input frequency. Its output is the adjustment that results from the filter frequency. According to embodiments of the present invention, the memory is further used to store a gain value that is selected based on a frequency input from a user. Accordingly, level compensation is performed based on the selected frequency, as described in detail above.
[0064]
In addition to the need to compensate for each control that produces a level change separately, there are situations where this solution is not sufficient. In a predictable situation (the combination of sound source and equalize type is consistent), the above process yields very good results. The more limited the type of sound source material and the more limited the range / effect of the equalizing element, the easier it is to develop a empirically derived compensated processor. However, the more diverse the sound sources, the more tests are required, and eventually the test results become inconsistent.
[0065]
The most complex situation is that for all possible human listeners (after allowing each person to be different because the Fletcher Manson curve is an average value) For speech, music, sound effects, etc.) one would design with a knob having a wide range (eg frequency control in the range of 20-20 kHz). This is not possible in practice because changes in the sound source interfere with a given equalizer. For example, assuming a bass boost frequency knob that works well for a wide range of sound sources such as orchestras, the compensation circuit amplification level increases at low frequencies (this is the Fletcher Manson effect) This is likely to be done for It is assumed that there is a music part where only the music triangle is played. When adjusting the equalizer knob at low frequencies, the triangle has no low frequency that can be emphasized, so there is no effect on the tone of the triangle. Although the change is not audible at all, the gain compensator continues to add levels, so turning the knob back and forth will change the volume of the triangle. In the case of a triangle (or any audio that does not have a low frequency), this knob turns into volume control and confuses the user. That is, a circuit that compensates only based on the setting position of the equalizing knob (as determined by any method, including those described above) will not work well in significantly different situations. In such a situation, the first embodiment described above is preferred.
[0066]
A third embodiment of the present invention is shown in FIGS. 13 to 16 as a dynamic processing compensation circuit. A dynamic processor can achieve any number of effects, such as limiting peaks to prevent overloading, compressing the dynamic range of a signal, expanding the range, or keeping a note long, etc. Change the signal gain. Since the total gain varies based on the input signal level, it is very difficult to compensate for parameter adjustments using the source-adjusted compensation method described above. Such a method tends to cancel the work of the processor. Instead, according to an embodiment of the present invention, an expected gain change can be compensated by the following procedure.
1. Assume the normalized listening level desired after processing. The simplest normal point is 0 dBV.
2. Without compression with a gain of 1, a 0 dBV input signal will produce a 0 dBV output, so the gain change from compression of a 0 dBV input signal will be matched by an equal and opposite gain change to maintain a 0 dBV output. Should. For example, if it is desired to reduce the dynamic range from 90 dB to 60 dB, a compressor is used to gradually reduce the gain as the input signal rises to 0 dBV, starting from -90 dBV. However, the output of a 0 dBV input will be -30 dBV. It is more likely that the range of −60 to 0 dBV is desirable rather than −90 to −30 dBV. Therefore, the gain should be increased by 30 dB.
3. The gain adjustment is shown in FIG. Using a voltage controlled amplifier (VCA) with positive and negative control voltage inputs, a constant control voltage equal to the control voltage expected at the other terminal with 0 dBV input is supplied to one terminal. For example, a 0 dBV input signal at a certain parameter setting generates +100 mV at the negative control port to reduce the gain. The +100 mV voltage at the positive control port negates this reduction. Also, equal and opposite control voltages may be coupled at a single terminal of the same circuit.
[0067]
To implement this method, the same circuit is used to modify both the RMS output signal and a constant voltage source equal to the RMS output at the normalized input level (0 dBV in this case). In this embodiment, a dual potentiometer is used for the threshold and ratio knobs and both control voltage circuits can be adjusted simultaneously, so the two control voltages change in tandem. Other circuits can also be used to accomplish the same function.
[0068]
Referring to FIG. 14, the transfer function V of a normal compressor _in V _out The graph is shown. Note how much the peak level drops for low thresholds and large ratios. FIG. 15 shows a graph of a level compensated compressor at various settings with the level normalized to 0 dBV (FIG. 15A shows the same effect normalized to 14 dBV). For example, at a 2: 1 ratio, compression of the -16 dBV threshold will reduce gain by -8 dB at 0 dBV input. Therefore, the gain is compensated by +8 dB. At ∞: 1, the gain must be compensated +16 dB. It should be noted that normalizing at a high level of dynamic range will increase the volume of most signals that are typically below 0 dBV. However, this gain adjustment is usually desirable unless the goal is simply to “prune up”. In FIG. 16, a detailed circuit diagram of a dynamic processor using a THAT 4301 dynamic processor chip is shown integrated with an RMS converter and a log-based VCA.
[0069]
Embodiments of the present invention compensate for level differences caused by changes in the signal processor. The compensator should only modify the level based on the modification of the processor signal. To avoid false activation of level compensation, the detection circuit or switch should freeze the level compensator from its modification changes when the signal processor is not tuned.
[Brief description of the drawings]
FIG. 1 is an overall configuration diagram of an apparatus configured according to an embodiment of the present invention.
FIG. 2 is an overall configuration diagram of a second embodiment of the present invention.
FIG. 3 is a detailed circuit diagram of the processor of FIG. 2;
4 is a detailed circuit diagram of the amplifier of FIG.
FIG. 5 is a graph showing the relationship between frequency and level compensation in the example of operation of FIGS.
FIG. 5A is a notch filter with interworking depth (depth) and frequency level compensation constructed in accordance with an embodiment of the present invention.
FIG. 6 is a block diagram of a level compensator configured by an embodiment of the present invention in which a single element of a potentiometer can change both frequency and gain.
7 is a detailed circuit diagram of the apparatus of FIG.
FIG. 8 is a block diagram of a digital level compensation system configured in accordance with an embodiment of the present invention.
FIG. 9 is a block diagram of an apparatus configured according to the first embodiment of the present invention.
FIG. 10 is a detailed example of the apparatus of FIG.
11 is a more detailed circuit diagram of some elements of FIG.
12 is a more detailed circuit diagram of some elements of FIG.
FIG. 13 is a configuration diagram of a third embodiment of the present invention.
FIG. 14 is a graph showing the effect of compression well known in the art.
FIG. 15 is a graph illustrating the effect of compression with level compensation according to an embodiment of the present invention.
FIG. 15A is a graph similar to FIG. 15 normalized to different levels.
FIG. 16 is a detailed circuit diagram of a dynamic processor that implements a third embodiment of the present invention.

Claims (19)

The chosen instrument as an input signal, at least one frequency band, the level compensation to compensate for changes in signal level caused by the signal processor from the first signal level to said input signal to a second signal level for the modified signal A system comprising:
A comparison unit adapted to be coupled to the output of the signal processor and adapted to receive the input signal and the modified signal from the signal processor, the comparison unit comprising the input signal and the A comparison unit adapted to determine a difference in signal level between the corrected signal;
A level compensation unit coupled to the output of the comparison unit and adapted to be coupled to the output of the signal processor, the level compensation unit being configured to detect the signal of the modified signal based on the difference in the signal level. A level compensation unit adapted to modify the level.   The level compensation system according to claim 1, wherein the input signal is an audio signal. further,
A signal processor coupled to the comparison unit and the level compensation unit, the signal processor comprising an equalizer, a vocoder, a distortion effect component, a chorus effect component, a flanger effect component, a ring modulator, and a wah wah effect component And a signal processor including at least one of a compressor, an expander, a tremolo component, a vibrato component, a reverberation component, and a delay component. The chosen instrument as an input signal, at least one frequency band, the level compensation to compensate for changes in signal level caused by the signal processor from the first signal level to said input signal to a second signal level for the modified signal A system comprising:
A comparison unit adapted to receive the input signal;
A level compensation unit adapted to be coupled to an output of the signal processor and an output of the comparison unit, the level compensation unit modifying the signal level of the modified signal based on the output of the comparison unit A level compensation unit adapted to
A level compensation system, wherein the comparison unit is adapted to output a signal indicative of a signal level difference between the input signal and a signal output by the level compensation unit.   The level compensation system according to claim 4, wherein the input signal is an audio signal. further,
A signal processor coupled to the comparison unit and the level compensation unit, the signal processor comprising an equalizer, a vocoder, a distortion effect component, a chorus effect component, a flanger effect component, a ring modulator, and a wah wah effect component 6. The level compensation system of claim 5, comprising a signal processor including at least one of: a compressor, an expander, a tremolo component, a vibrato component, a reverberation component and a delay component. The comparison unit further comprises:
A first root mean squared converter adapted to receive the input signal and measure a root mean square level of the input signal ;
A second root mean squared converter adapted to receive the output signal from the level compensation unit and measure a root mean square level of the output signal ;
5. The level compensation system of claim 4, comprising a comparator coupled to the first and second root mean squared transducers. The level compensation unit further comprises:
A circuit including a digitally controlled resistor having a first input coupled to the comparison unit and a second input adapted to receive the modified signal, the circuit based on the output of the comparison unit The level compensation system of claim 4, further comprising a circuit adapted to modify a volume level of the modified signal. A method for performing level compensation on an input signal,
Processing the input signal in a signal processor to generate a modified signal;
Comparing the signal level of the input signal with the signal level of the modified signal;
Modifying the signal level of the modified signal based on the comparing step.   The method of claim 9, wherein the comparing step calculates a signal level difference between the input signal and the modified signal.   The method of claim 10, wherein in the modifying step, the signal level of the modified signal is modified to match the signal level of the input signal. A level compensation method,
(A) supplying the selected sound source as an input signal to the signal processor;
(B) adjusting the control of the signal processor by a predetermined amount;
(C) comparing the level of the input signal with the level of the modified signal output by the signal processor;
(D) determining a difference value representing a level difference between the input signal and the modified signal;
(E) repeating steps (a) to (d) for a plurality of input signals;
(F) calculating an expected difference in signal level generated by adjusting the control of the signal processor based on the difference value determined in step (d). The method of claim 12, wherein step (b) comprises adjusting the control multiple times .
(A ) supplying the selected sound source as an input signal to the signal processor;
(B) adjusting the control of the signal processor by a predetermined amount;
(C) comparing the level of the input signal with the level of the modified signal output by the signal processor;
(D) determining a difference value representing a level difference between the input signal and the modified signal;
(E) repeating steps (a) to (d) for a plurality of input signals;
(F) calculating an expected difference in signal level generated by adjusting the control of the signal processor based on the difference value determined in step (d). further,
The method of claim 12, comprising (g) modifying the signal level of the modified signal based on the expected difference in the signal level calculated in step (f). A level compensation method,
(A) providing an input signal to a dynamic signal processor, wherein the dynamic signal processor depends on a predetermined ratio when the input signal has a signal level that intersects a predetermined threshold; Being adapted to change the gain of the signal,
(B) providing an output control voltage signal from the dynamic signal processor to an amplifier that causes a change to the level of the input signal;
(C) supplying a constant control voltage equal to an expected output control voltage signal from the dynamic signal processor of the input signal normalized based on the setting of the predetermined ratio and the predetermined threshold; When,
(D) modifying the input signal in the amplifier based on a difference between the output control voltage and the constant control voltage. A circuit for controlling the gain and frequency correction of an input signal,
A filter circuit adapted to receive the input signal and to modify a signal level of the input signal in at least one frequency range, wherein the frequency range is selected based on a control input;
A gain circuit adapted to receive the modified signal from the filter circuit and the signal from the control input, the gain circuit being configured to receive the modified signal based on the signal from the control input. A gain circuit adapted to modify the gain.   The circuit of claim 16, wherein the control input selects a resistance value of each of the filter and the gain circuit.   The circuit of claim 17 wherein the control input is part of a potentiometer.   The circuit of claim 18, wherein the control input is generated from a single control element of the potentiometer.

Images