DE102015004520B4 - Method for the automatic generation of an accompaniment consisting of tones and device for its execution - Google Patents

Abstract

A method for automatically generating an audio accompaniment by a signal processing device using a sound signal comprising the steps of: - starting a tracking frequency determination when a trigger signal exceeds a trigger threshold, the tracking frequency determination comprising the steps of: - performing a Fourier transform on Sound signal or a signal derived from the sound signal, so that a frequency spectrum is obtained, - analyzing the frequency spectrum by means of a predetermined analysis scheme, so that an analysis result is obtained, - determining a concomitant frequency as a function of the analysis result, - outputting an accompaniment tone, the fundamental frequency of the accompanying frequency is.

Description

The invention relates to a method for automatically generating an accompaniment consisting of tones according to claim 1 and to an apparatus for carrying out this method according to claim 9.

Especially with musicians who perform alone, they often have the desire to accompany themselves with another instrument. For this purpose, in particular the case of a solo guitarist is mentioned, especially since the most important application case of the invention described below is the generation of an accompaniment starting from a guitar playing.

For rhythmic self-accompaniment, especially of guitarists, so-called stomp boxes are known, on which the guitarist "beats" the beat (and possibly also a slightly more complicated rhythm) with his foot. If the guitarist also wants a tonal accompaniment (especially a bass accompaniment), so far he has only the opportunity to play a previously completely stored accompaniment (nowadays usually of a digital medium) and then the main instrument (ie the guitar ) to play. However, this ultimately turns the principle of live performance on its head, as the live instrument (the guitar) must follow the accompaniment instrument (especially the previously recorded bass). As a result, an essential aspect of live playing is lost and the guitarist has no room for improvisation or spontaneous tempo changes.

The WO 2014/086935 A1 describes a method for generating an accompaniment in which a recognition module first classifies a piece of music fed into it. Starting from this classification, a selection unit selects a previously recorded piece of music and this is output via an output.

Also the US 4,922,797 A describes a system for creating a kind of accompaniment, but the musician can intervene in the production of the accompaniment.

In the US 2009/0064851 A1 For example, a method for generating an accompaniment of a previously recorded piece of music in which the previously recorded piece of music is subjected to a frequency analysis is described. Based on the frequency analysis of the entire piece of music, a suitable accompaniment is calculated.

From the DE 2 407 108 A is an electronic organ with automatic bass accompaniment known using the stop signals of the keys.

On this basis, the present invention has the object to provide a method for the automatic generation of a sound consisting of accompaniment by a signal processing device using a sound signal, which allows an instrumentalist, in particular a guitarist, himself with another instrument, in particular a Bass, to accompany, but the accompaniment but not previously recorded and thus must be fixed.

This object is achieved by a method having the features of claim 1. An apparatus for carrying out the method is specified in claim 9.

For each chord played, there is at least one accompaniment tone always fitting, in particular the root note of the chord. To name a concrete example, this is an a minor for an a minor chord. Therefore, the basic idea of the invention is to analyze a played chord of a sound signal by means of an electronic signal processing device at a certain time, so as to recognize the type of chord and automatically generate from this information an accompaniment tone with a suitable accompanying frequency. For this purpose, the method has the following steps:
When a trigger signal exceeds a trigger threshold, a tracking frequency decision is started which has several steps. First of all, a Fourier transformation is performed on the audio signal or on a signal derived from the audio signal (as a rule this is the digitized audio signal), so that a frequency spectrum is obtained. In the next step, the obtained frequency spectrum is analyzed by means of a predetermined analysis scheme, so that an analysis result is obtained. This analysis result indicates which chord the audio signal contains at the time of performing the Fourier transformation (this is the trigger time up to a small delay of a maximum of a few ms), or at least makes a statement about the fundamental tone of this chord. Then, the method determines an accompanying frequency depending on the result of the analysis. In most cases, the corresponding assignment rule will be very simple, stating that the accompaniment frequency is the frequency of the fundamental of the detected chord, or one or more octaves away from it. In the case of bass accompaniment, the rule may be, in particular, that the accompaniment frequency is half the frequency of the fundamental tone of the determined chord. Finally, an accompaniment tone whose fundamental frequency is the accompaniment frequency is output.

In order to give the musician the greatest possible freedom, at which time the accompaniment sound is generated, the trigger signal in a preferred embodiment is a rhythm signal different from the sound signal generated by the musician - for example by means of a so-called stomp box - and the signal processing device a signal line is supplied.

There are basically several analysis schemes with which the frequency spectrum obtained by the Fourier transformation can be analyzed. Which scheme is chosen depends on several factors, in particular on the available computing power, on the required recognition accuracy and on whether presets are to be made with regard to the chords to be recognized or not. This will be dealt with separately later. A first possible analysis scheme is given in claim 4.

Another preferred analysis scheme consists of a pattern comparison. Here, the measured frequency spectrum is compared with a number of stored reference spectra, wherein each of the stored reference spectra is uniquely associated with a chord, so that it is also assigned a reference frequency. To stay with the above example: the stored reference spectrum of an a minor chord is assigned the reference frequency of an a. By pattern comparison, the measured frequency spectrum is assigned the correct reference spectrum and thus the reference frequency is determined. The accompanying frequency is then usually the reference frequency (frequency of the fundamental of the determined reference spectrum) or one or more octaves spaced therefrom.

In order to minimize the computational effort on the one hand and the error rate on the other hand, it is preferred that the key in which a piece is played, is preselected, so that each measured frequency spectrum is compared only with the reference spectra belonging to the preselected key. As a result, the distance between the individual reference spectra is doubled, which of course simplifies the pattern comparison.

To achieve a natural sound result, the accompaniment sounds are preferably samples, for example of a bass.

The invention will now be explained in more detail using an exemplary embodiment with reference to the figures. Hereby show:

1 a highly schematic representation of an embodiment of an apparatus for automatically generating an accompaniment,

2 a flow chart of a preferred method for automatically generating an accompaniment, as for example by the device of 1 can be carried out,

3 a schematic representation of a measured frequency spectrum and

4a -C schematic representations of three reference spectra with which the measured frequency spectrum is compared.

The inventive method can in principle be performed by any computer. However, since the method is intended in particular for use on stage, the method is usually carried out by a device specially manufactured for this purpose. A corresponding device is under the reference numeral 10 in 1 shown schematically. The device 10 has a housing 12 on, at which a first signal input 13 , a second signal input 14 and a signal output 15 located. The inputs and the output are usually designed in the form of sockets for jack plugs. There is also a key selector 18 provided by means of which the played key can be selected.

The just mentioned elements are with a signal processing device 20 connected, which is inside the housing 12 located. This has one with the first signal input 13 connected first analog-to-digital converter 22 , one with the second signal input 14 connected second analog-to-digital converter 24 and one with the signal output 15 connected digital-to-analog converter 36 on. Whether the analog-to-digital converters and the digital-to-analog converter are considered as belonging to the signal processing device, or only to its periphery, is a definition problem that is irrelevant to the invention. However, it is important to note that also embodiments of the device are conceivable, which do not require A / D and D / A converter. This would be possible if the incoming and outgoing signals are digital signals, which is conceivable.

The actual data processing unit has a processor 30 , a bi-directional RAM connected to the processor 32 and a ROM memory 34 on.

The two analog-to-digital converters 22 . 24 and the key selector 18 give their signals to the processor 30 which in turn sends signals to the digital-to-analog converter 36 passes.

In ROM memory 34 In particular, reference spectra and samples for the accompaniment tones are stored permanently.

Of course, a power supply (via battery or power supply) is provided, but this is not shown.

The method used by the device just described 10 will be performed with reference to the 1 . 2 . 3 and 4a to 4c explains:
To the first signal input 13 an instrument is connected by means of cable or radio connection to which an accompaniment is to be generated. This instrument can be a guitar in particular. To the second signal input 14 a trigger transmitter is connected, which can be designed in particular as a so-called stomp box.

At the beginning, the musician selects by means of the key selector switch 18 a key from, in a preferred embodiment, the twelve conventional major and minor keys can be selected.

After selecting the key, the musician can begin to play on his guitars, with the sound signal of the guitar over the first signal input 13 the device 10 supplied and first by means of the first analog-to-digital converter 22 is digitized. The then digitized sound signal is the processor 30 fed. Furthermore, to the second signal input 14 a rhythm instrument, in particular a stomp box connected, which supplies a rhythm signal to the device, which serves as a trigger signal. Also, this trigger signal is first digitized, namely by means of the second analog-to-digital converter 24 so that the processor 30 a digitized trigger signal is supplied. Because the processor 30 Of course, it is only possible to process digitized signals. For the purpose of linguistic simplification, only the signals (not the digitized signals) will be discussed below.

As long as the trigger signal does not exceed a pre-set trigger threshold, the system is in the wait state. If the musician now steps his foot on the stomp box, the trigger signal exceeds a set trigger threshold and the process of setting the reference frequency is started. The process of Accompany Frequency Determination begins with the caching of the current audio signal in RAM 32 , At this stored sound signal, a Fourier transformation is now performed, so that a frequency spectrum is obtained. Such a frequency spectrum is symbolic in 3 It should be emphasized, of course, that the corresponding frequency spectrum exists only in the form of digital information and not graphic information. However, as it helps to understand the process, a graphically rendered spectrum is shown. If a chord is played on a guitar so that all six sides are struck, then the frequency spectrum has six peaks, as in 3 also shown. The frequency spacings are chosen arbitrarily here and do not necessarily correspond to an actual harmonic chord. Of course, only the peaks, that is, their peak frequency and their peak amplitude, are interesting for the analysis to be performed 3 also shown noise is usually filtered out.

The obtained frequency spectrum of 3 is now analyzed on the basis of a given analysis scheme, the preferred and explained here analysis scheme on a pattern comparison with reference spectra, which in the ROM memory 34 are stored. Each selected key usually belongs 9 Chords, so nine reference spectra. Each of these reference spectra is assigned the root of its chord, the reference spectrum of an a minor chord, so for example an a. The measured spectrum ( 3 ) is now compared to the reference spectra. For the sake of clarity are in the 4a to c only three, not nine reference spectra shown, which is sufficient for the explanation, however. Now, the reference spectrum suitable for the measured spectrum is selected, this would be the reference spectrum of the 4b , Thus, the reference frequency is determined, namely the reference frequency of the determined reference spectrum.

The processor 30 now reads from the ROM memory 34 a sample with a matching frequency matching the reference frequency, wherein the accompanying frequency in particular have the same fundamental frequency as the reference frequency or one or more octaves can be spaced from this. In the case of a bass accompaniment, the procedure is usually such that the accompaniment frequency is one octave lower than the reference frequency. The selected sample is from the processor 30 now the digital-to-analog converter 36 which supplies the electrical analog signal of the accompaniment tone to the signal output 15 passes.

The system now waits until the next time the trigger threshold is exceeded, then the process just described begins again.

LIST OF REFERENCE NUMBERS

10

Apparatus for generating an accompaniment

12

casing

13

first signal input

14

second signal input

15

output

18

Key selector

20

Signal processing device

22

first analog-to-digital converter

24

second analog-to-digital converter

30

processor

32

RAM

34

ROM memory

36

Digital to analog converter

40

signal output

Claims (10)

A method for automatically generating an audio accompaniment by a signal processor using a sound signal, comprising the steps of: - starting a tracking frequency determination when a trigger signal exceeds a trigger threshold, wherein the tracking frequency setting comprises the steps of: Performing a Fourier transformation on the audio signal or on a signal derived from the audio signal, so that a frequency spectrum is obtained, Analyzing the frequency spectrum using a predetermined analysis scheme so that an analysis result is obtained, - determining an accompanying frequency as a function of the analysis result, - Outputting an accompaniment tone whose fundamental frequency is the accompanying frequency. A method according to claim 1, characterized in that the trigger signal is a rhythm signal different from the audio signal. A method according to claim 2, characterized in that the rhythm signal is supplied from an external source of the signal processing means. Method according to one of the preceding claims, characterized in that the analysis scheme comprises the following steps: - selecting a predetermined number of the highest amplitude peaks, - selecting the two lowest frequency peaks having the frequency spacing of a fifth, - selecting the peak frequency of the lower frequency of the two peaks, - Set the tracking frequency as a function of the selected peak frequency. A method according to claim 4, characterized in that the accompanying frequency is the frequency of the selected peak frequency or one or more octaves is spaced therefrom. Method according to one of claims 1 to 3, characterized in that the analysis scheme comprises the steps of: - comparing the measured frequency spectrum with a number of stored reference frequency spectra, wherein each reference frequency spectrum is assigned a reference frequency, - selecting the reference frequency spectrum which most closely resembles the measured frequency spectrum is, - determining the accompanying frequency, wherein the accompanying frequency is the reference frequency of the selected reference spectrum or one or more octaves is spaced therefrom. A method according to claim 6, characterized in that a key is preselected and that the measured frequency spectrum is compared only with the reference spectra belonging to the preselected key. Method according to one of the preceding claims, characterized in that the accompaniment sound is a sample. Device for carrying out a method according to one of Claims 2 to 8, having a first signal input ( 13 ) for supplying the audio signal, a second signal input ( 14 ) for supplying the rhythm signal, a signal processing device connected to the two signal inputs ( 20 ) for generating accompaniment tones and one with the signal processing device ( 20 ) connected signal output ( 15 ) for outputting the accompaniment tones, wherein the signal processing device ( 20 ) is arranged to perform accompaniment frequency determinations and to generate accompaniment tones according to any one of claims 1 to 8. Apparatus according to claim 9, characterized in that it further comprises a preselector for preselecting a key.

Images