WO2023156578A1

WO2023156578A1 - Method for processing a digital sound signal for vinyl disc emulation

Info

Publication number: WO2023156578A1
Application number: PCT/EP2023/054001
Authority: WO
Inventors: Franck Rosset; Jean-Luc HAURAIS
Original assignee: Franck Rosset; Haurais Jean Luc
Priority date: 2022-02-21
Filing date: 2023-02-17
Publication date: 2023-08-24
Also published as: FR3132974A1

Abstract

The present invention relates to the field of digital processing in order to reproduce the auditory impression of a "vinyl sound". Vinyl records were created in 1948; the sound information was reproduced by the movement of a needle transmitting the deviations in the groove to an electromagnetic or piezoelectric transducer that converts said vibrations into an electrical signal. When manufacturing the disc matrix, high frequencies are boosted and low frequencies are reduced according to an RIAA curve. This enables a more regular engraving dimension to be achieved and prevents high frequencies from being lost in background noise and low frequencies from sending the engraving point into the neighbouring turn. Playback is corrected according to the reverse RIAA curve.

Description

METHOD FOR PROCESSING A DIGITAL SOUND SIGNAL FOR EMULATION OF VINYL RECORDS

Field of invention

The present invention relates to the field of digital processing to restore the auditory impression of a “vinyl sound” from a reproduction on digital audio equipment.

The vinyl microgroove record was born in 1948; the sound information was restored by the movement of a needle transmitting the deviations of the groove to an electromagnetic or piezoelectric transducer which transforms these vibrations into an electrical signal. When manufacturing the disk matrix, the high frequencies are boosted and the lows are cut according to an RIAA curve. This makes it possible to have a more regular engraving dimension and prevents the highs from getting lost in the background noise and the lows from sending the chisel into the neighboring turn. Reading is performed with a correction according to the inverse curve RIAA.

The imperfections resulting from this "mechanical" reading (wear, breath, surface noise, speed fluctuations, crackles, ...) led in the 2000s to favor digital recordings on physical media (CD, DVD-audio) or dematerialized (digital radio, DAB, "streaming").

Recently, there has been a return of interest in vinyl records, not just out of nostalgia, but because of a more musical perception of listening. This difference in perception is explained by the specificities of the recording of vinyl records, which in the absence of sampling and because of the "physical" constraints imposed by the formation of the grooves, and resulting in particular in typical parameters ( RIAA curve), which bring a particular "coloration" of an easier to listen to signal, in which more details are brought to audible thresholds and by a sound perceived as more musical, warmer and more "punchy".

In order to be able to reconcile the advantages of a digital signal, a priori unalterable, and the particularity of "vinyl sound", to use this term designating a somewhat subliminal perception and difficult to characterize technically, it has been proposed in the prior art to apply filters to change the spectral curve of the sound signal and simulate "vinyl sound" from a digital signal.

State of the art

Known in the state of the art is the patent application US2011060436 describing a stereo biaural audio recording method, in which the notes stored in a first channel of the recording are stored in the same sequence in a second channel of the recording but are shifted in frequency by the same value, and on a method for processing an audio signal, consisting in:

- acquiring a set of data capable of defining an audio signal;

- select a frequency offset;

- generating a first digital audio recording by synthesizing it according to the data set;

- generating a second digital audio recording by synthesizing it according to the data set and the frequency offset;

- select a channel from the first audio recording, select a channel from the second audio recording, mix the two channels into a third audio recording which includes at least two channels.

We also know the article Jot, Jean-Marc & Walsh, Martin. (2010). Center-channel processing in virtual 3-D audio reproduction over headphones or loudspeakers. 128th Audio Engineering Society Convention 2010. 3. 1935-1944 concerning virtual 3D audio processing systems intended for the spatial enhancement of recordings reproduced through headphones or loudspeakers generally produce a less convincing effect on the sound components located at the center. The virtual 3D audio processing algorithm described makes it possible to improve the spatial reinforcement effect in the headphones or in the loudspeakers, even for the sound components located in the center of the stereo image, to preserve the timbre and the balance. of the original recording and produce a more stable "ghost center" image in the loudspeakers. The proposed improvements are beneficial, in particular, in the audio systems of computers and televisions or in the speakers of the "soundbar" of the home theater.

l'accès à une paire de réponses impulsionnelles acoustiques binaurales (BAIR) individualisée pour un auditeur dans une position de référence ;
l'accès à une pluralité de réponses impulsionnelles acoustiques binaurales (BAIR) pour l'auditeur correspondant à des positions supplémentaires par rapport à l'auditeur ;
la dérivation d'une pluralité de fonctions de transfert pour convertir des réponses impulsionnelles acoustiques binaurales (BAIR) pour la position de référence par rapport à l'auditeur à chacune des positions supplémentaires en divisant chacune de la pluralité de paires de réponses impulsionnelles acoustiques binaurales (BAIR) pour les positions supplémentaires par la paire de réponses impulsionnelles acoustiques binaurales (BAIR) pour la position de référence;
la réception d'un signal indiquant une modification de l'orientation de la tête, et la sélection d'une paire de la pluralité de fonctions de transfert en réponse au signal et de façon correspondante au signal ; et
l'application de la paire de réponses impulsionnelles acoustiques binaurales (BAIR) de la position de référence et de la paire sélectionnée de la pluralité de fonctions de transfert au signal audio d'entrée pour localiser l'audio dans les écouteurs.

Patent application US2019379995 describes a method of providing a head-related transfer function (HRTF) to be applied to an input audio signal for audio localization in headphones, comprising:

accessing a binaural acoustic impulse response (BAIR) pair individualized for a listener in a reference position;
accessing a plurality of binaural acoustic impulse responses (BAIR) for the listener corresponding to additional positions relative to the listener;
deriving a plurality of transfer functions to convert binaural acoustic impulse responses (BAIR) for the reference position relative to the listener at each of the additional positions by dividing each of the plurality of pairs of binaural acoustic impulse responses ( BAIR) for the additional positions by the pair of binaural acoustic impulse responses (BAIR) for the reference position;
receiving a signal indicating a change in the orientation of the head, and selecting a pair of the plurality of transfer functions in response to the signal and corresponding to the signal; And
applying the pair of binaural acoustic impulse responses (BAIR) of the reference position and the selected pair of the plurality of transfer functions to the input audio signal to locate the audio in the headphones.

The article “Pitch extraction and separation of overlapping speech” P.N. Denbigh, J. Zhao, Speech Communication,Volume 11, Issues 2–3, 1992,Pages 119-125, ISSN 0167-6393 describes algorithms which separate two signals which overlap. These algorithms depend on accurate measurements of the pitch of the target voice. The first algorithm uses a single microphone and the important feature is to exploit the onset of voicing to extract its pitch in the presence of interference. The second algorithm uses two microphones and its major feature is also to exploit the direction of the target voice.

la transcription de la musique,
l’analyse des performances musicales, et
l'identification et la catégorisation de la musique.

The article “Audio content analysis” by Alexander Lerch et al. doi={10.48550/ARXIV.2101.00132 introduces the general audio content analysis process and examines basic approaches to audio content analysis. The different tasks of audio content analysis are classified into three categories:

music transcription,
musical performance analysis, and
identification and categorization of music.

Examples of music transcription systems include musical key detection, fundamental frequency detection, and musical pattern detection. Musical performance analysis systems present an overview of approaches to rhythm and tempo detection as well as musical performance evaluation. The music classification systems covered are audio fingerprinting, musical genre classification and musical emotion recognition. The chapter ends with a discussion of the current challenges in the field and a speculation on the prospects.

Disadvantages of the prior art

The prior art solutions result in a sound that is closer to a caricature of vinyl sound, adding to the flaws of digital sound the admittedly characteristic but nonetheless detrimental flaws of vinyl records. Unsophisticated listeners will of course find the special atmosphere of microgroove records.

Purists, for their part, will find with the solutions of the prior art only partially the acoustic specificities specific to vinyl sound.

Solution provided by the invention

In order to remedy these drawbacks, the present invention relates, in its most general sense, to a process for processing a pair of stereophonic signals having the characteristics set out in claim 1.

Une étape initiale de lecture sur une platine d’un disque microsillon vinyle physique, d’échantillonnage du couple de signaux stéréophonique et d’enregistrement lesdits signaux numérique stéréophoniques constituant les signaux stéréophoniques de référence, puis un traitement comportant :

It includes the following steps:

An initial step of playing a physical vinyl microgroove record on a turntable, of sampling the pair of stereophonic signals and of recording said digital stereophonic signals constituting the stereophonic reference signals, then processing comprising:

a) An iterative modification step consisting in applying a change of tone to increase the pitch for said initially reference stereophonic signals for the first modification, then to the pair of signals obtained during the previous iteration,

b) an additional step consisting of summing said digital stereophonic signals resulting from the last iteration and said reference stereophonic digital signals to produce a pair of corrected digital stereophonic signals.

Preferably, the number of iterations is between 3 and 8.

Advantageously, the method includes an additional step of reducing the stereophonic image of said pair of corrected digital stereophonic signals.

According to a variant, said additional step of reducing the stereophonic image consists in adding to each channel of said corrected digital stereophonic signals a central monophonic signal resulting from the processing of the two channels of said corrected digital stereophonic signals.

Advantageously, said processing of said corrected digital stereophonic signals consists of adding to the signal of each of the channels the opposite signal of the other channel and of adding the two resulting signals to provide a central monophonic signal.

According to a preferred embodiment, said initial step of playing a vinyl record is repeated with a plurality of turntables, and said pair of stereophonic reference signals corresponds to the selection of the pair of signals corresponding to one of said turntables as a function of a selection criterion based on characteristics specific to said plates.

According to another variant, the method comprises an additional step consisting in applying a post-processing of said pair of digital stereophonic signals corrected by an impulse response filter, by an impulse response corresponding to a capture of an impulse signal from an acoustic set and electronics for playing a vinyl record.

According to another variant, the method comprises an additional step consisting in generating an imprint of a reference listening system [that on the reference plate] in stereo and binaural at a distance of between 20 and 300 cm and in applying a filter to ensure maximum compatibility with HRTF profiles.

According to a particular mode of implementation, the method comprises processing by a crosstalk filter having an angle greater than 3 degrees and less than 320 degrees.

Advantageously, the method includes an additional step consisting in applying an equalization processing of the listening medium in order to restore a signal identical to the original.

According to another variant, the method comprises an additional step consisting in applying a specific processing to frequencies between 150hz and 300hz by impulse filtering with binaural impulse responses comprising a modification of delays of 25%, by application to each corrected signal of a low-pass filter and weighted addition of said corrected signal and of the signal processed by pulse filtering.

Advantageously, said processing comprises a modification of a value of between 10 and 20% of the "left" component of the "right" channel the impulse response as well as of the "right" component of the "left" channel the impulse response.

According to another variant, said method comprises an additional step consisting in applying a specific processing to the frequencies comprised between 12000 Hz and 22000 Hz with a specific processing to the harmonics comprised between 2 and 7 without modification of the harmonic 1, by application to each corrected signal d a high-pass filter and weighted addition of said corrected signal and of the processed signal.

According to a particular embodiment, the method comprises an additional step consisting in applying a processing of frequencies below 150 Hz in mono via a reduction of the stereo tracks with application of an acoustic filter.

According to another variant, the method comprises an additional step consisting in applying a multiband 3D acoustic correction filter defined by the pickup distance from the reference system and the volume of the harmonics.

According to another variant, it comprises an additional step consisting in applying a 3D acoustic correction filter applied to the fundamental.

Advantageously, the method includes an additional step of modifying the stereophonic image applied to step b) and to the fundamental of the signals corrected on the 3D signal.

Detailed description of a non-limiting embodiment of the invention.

La représente une représentation graphique d’un algorithme mis en œuvre du procédé objet de l’invention.

Other characteristics and advantages will emerge from the following description of the invention, description given by way of example only, with reference to the appended drawings in which:

There represents a graphical representation of an implemented algorithm of the method that is the subject of the invention.

General principles

Image stéréophonique limitée par l’extension physique maximale des sillons
Déformation : réglage de la quantité de déformation et la forme de déformation du disque, de l'absence de déformation aux bords totalement fondus et déformés.
Poussière : simulation de la quantité de poussière qui s'est déposée sur la surface du disque.
Année : Modèles de lecteurs de disques de différentes décennies en utilisant des réponses de filtre.
Usure : simulation de l'effet d'un disque qui a été joué trop souvent, du tout nouveau à quelques milliers de tours.
Bruit mécanique : ajout du grondement de la platine et le bruit du moteur.
Spin Down : simulation du son de l'arrêt lent de la lecture d'un enregistrement, en modulant à la fois la vitesse et la fréquence de lecture.

The object of the invention is to post-process a digital audio file to enable listening through headphones, with "modern" equipment, preserving the impression of listening to a microgroove record, characterized by the characteristic signature of their defaults :

Stereophonic image limited by the maximum physical extension of the grooves
Warp: Adjust the amount of warp and warp shape of the disc, from no warp to fully faded and warped edges.
Dust: simulation of the amount of dust that has settled on the surface of the disc.
Year: Record player models from different decades using filter responses.
Wear: Simulation of the effect of a record that has been played too many times, from brand new to a few thousand rpm.
Mechanical Noise: Added turntable rumble and motor noise.
Spin Down: simulation of the sound of the slow stop of the playback of a recording, by modulating both the speed and the frequency of playback.

The objective is to obtain with headphones a listening consistent with a vinyl mastering on a reference system while keeping all the advantages of the support, as well as the restitution of the space of the two speakers (left and right) and the listener. , essential for an identical reproduction of an existing system on a hifi system.

Digital processing

The digital sound signal is subject to a set of processing described below and illustrated by the .

The first step (1) consists of reading a vinyl record (30) on a turntable (40) and sampling this pair of stereophonic signals to record a digital reference file (35) taking into account all the audio characteristics of playing a vinyl record on a turntable. This reference file (35) is recorded in a memory, during an initial step of the method.

Listening to a digital file

When listening to a digital file (45), a series of digital stereo signal processing is carried out. We start by creating a harmonic of the fundamental frequency of this signal. To create an artificial harmonic, the signal is duplicated (step 2) and a frequency shift is performed on the duplicated component in a ratio x2 to create a harmonic 1 (step 3). Thus, the 100 Hz frequencies are transformed into 200 Hz frequencies, without the duration of the musical sequence changing. This processing can be carried out by applications of the type "voicechanger" (trade name).

Several iterations of this harmonic creation step are carried out, then the reference signal (35) is added to the signal resulting from the last iteration (step 4) of processing of said digital file (45).

The number of iterations is determined empirically, according to the evolution of the signals of an iteration compared to the signals of the previous iteration. If the variation is imperceptible, the treatment is stopped. Typically, the optimum number of iterations is between 3 and 8, and generally five iterations lead to a good result.

This first processing of the digital signal (45) integrates the characteristics specific to vinyl records by this step of combining the digital signal (45) and the reference signal (35) recorded during the initial step.

Étape (5) : modification de l’image stéréophonique résultant de l’étape (4) et à la fondamentale des signaux corrigés sur le signal stéréophonique via un traitement dynamique défini par le contenu de la source audio
Étape (6) : Génération de l’empreinte d’un système Hifi de référence en stéréo et en binaurale à une distance comprise entre 20 et 300 cm.

We then proceed to a combination of one or more additional treatments:

Step (5): modification of the stereophonic image resulting from step (4) and at the fundamental of the corrected signals on the stereophonic signal via a dynamic processing defined by the content of the audio source
Step (6): Generation of the footprint of a reference Hifi system in stereo and binaural at a distance between 20 and 300 cm.

Étape (7) : Cette étape consiste à ajouter un filtre assurant une compatibilité maximale aux fonctions de transferts relatives à différents profils de tête HRTF (head-related transfer function)
Étape (8) ; Utilisation d’un filtre de diaphonie avec un angle supérieur à 3 degrés et inférieur à 320 degrés
Étape (9) : Égalisation du support d’écoute afin de restituer un signal identique à l’original
Étape (10) : Traitement spécifiques aux fréquences comprises entre 150hz et 300hz avec des réponses impulsionnelles binaurales comprenant une modification de délais de 25% ainsi qu’une modification du signal gauche de la réponse impulsionnelle droite ainsi que le signal droit de la réponse impulsionnelle gauche d’une valeur comprise entre 10 et 20%.
Étape (11) : Traitement spécifique aux fréquences comprises entre 12000hz et 22000hz avec un traitement spécifique aux harmoniques comprises entre 2 et 7 sans modification de l’harmonique 1
Étape (12) : Traitement des fréquences inférieures à 150hz en mono via une réduction des pistes stéréo avec application d’un filtre acoustique.
Étape (13) : Filtre de correction acoustique 3D multibandes définit par la distance de captation du système de référence et du volume des harmoniques
Étape (14) : filtre de correction acoustique 3D appliqué à la fondamentale
Étape (15) : modification de l’image stéréophonique appliquée à l’étape b) et à la fondamentale des signaux corrigés sur le signal 3D
Étape (16) : Utilisation des signaux de références afin de générer le bruit caractéristique du vinyle
Étape (17) : Traitement du signal issu de l’étape (16) avec un jeu de réponses impulsionnelles spacialisées spécifiques.

This step aims to produce a binaural audio signal using a sound capture method adapted to the morphology of the human head. It consists of recording from an artificial head (36) a reference signal reproduced by loudspeakers placed at a distance of between 20 and 300 centimeters to record an impulse response (37). This impulse response (37) will then be used for binaural reproduction in headphones.

Step (7): This step consists of adding a filter ensuring maximum compatibility with the transfer functions relating to different HRTF head-related transfer function profiles.
Step (8); Using a crosstalk filter with an angle greater than 3 degrees and less than 320 degrees
Step (9): Equalization of the listening medium in order to restore a signal identical to the original
Step (10): Specific processing at frequencies between 150hz and 300hz with binaural impulse responses including a 25% delay modification as well as a modification of the left signal of the right impulse response as well as the right signal of the left impulse response with a value between 10 and 20%.
Step (11): Specific processing for frequencies between 12000hz and 22000hz with specific processing for harmonics between 2 and 7 without modification of harmonic 1
Step (12): Treatment of frequencies below 150hz in mono via reduction of stereo tracks with application of an acoustic filter.
Step (13): Multiband 3D acoustic correction filter defined by the pickup distance from the reference system and the volume of the harmonics
Step (14): 3D Acoustic Correction filter applied to fundamental
Step (15): modification of the stereophonic image applied in step b) and to the fundamental of the signals corrected on the 3D signal
Step (16): Use of reference signals to generate the characteristic noise of vinyl
Step (17): Processing of the signal resulting from step (16) with a set of specific spatialized impulse responses.

The objective is to obtain with headphones (20) a listening consistent with a vinyl mastering on a reference system while keeping all the advantages of the support, as well as the restitution of the space of the two speakers (left and right) and of the listener, essential for an identical reproduction of an existing system on a hi-fi system.

Claims

Process for processing a pair of stereophonic signals characterized in that it comprises:
a) An initial reading step consists of reading a vinyl record (30) on a turntable (40) and sampling this pair of stereophonic signals to record a digital reference file (35) taking into counts all the audio characteristics of playing a vinyl record on a turntable, this reference file (35) being recorded in a memory, during an initial step of the method,
b.) Then, when listening to a digital file (45), processing comprising the following steps:
i. An iterative modification step consisting in applying a change of tone to increase the pitch for said initially reference stereophonic signals for the first modification, then to the pair of signals obtained during the previous iteration,
ii. an additional step consisting of summing said digital stereophonic signals resulting from the last iteration and said reference digital stereophonic signals to produce a pair of corrected digital stereophonic signals.
Process for processing a digital sound signal according to Claim 1, characterized in that the number of iterations is between 3 and 8.
Process for processing a digital sound signal according to Claim 1, characterized in that it includes an additional step of reducing the stereophonic image of said pair of corrected digital stereophonic signals.
Process for processing a digital sound signal according to the preceding claim, characterized in that the said additional step of reducing the stereophonic image consists of adding to each channel of the said corrected digital stereophonic signals a central monophonic signal resulting from the processing of the two channels of the said signals. corrected digital stereophonics.
Process for processing a digital sound signal according to the preceding claim, characterized in that the said processing of the said corrected digital stereophonic signals consists of adding to the signal of each of the channels the opposite signal of the other channel and of adding the two resulting signals to provide a central monophonic signal.
Process for processing a digital sound signal according to Claim 1, characterized in that the said initial stage of playing a vinyl record is repeated with a plurality of turntables, and in that the said pair of reference stereophonic signals corresponds to the selection of the pair of signals corresponding to one of said plates as a function of a selection criterion based on characteristics specific to said plates.
Process for processing a digital sound signal according to Claim 1 or 2, characterized in that it comprises an additional step consisting in

recording from an artificial head (36) a reference signal reproduced by loudspeakers placed at a distance of between 20 and 300 centimeters to record an impulse response (37),

this impulse response (37) then being used for binaural reproduction in headphones, by a processing operation comprising the following steps:

Step (7): Addition of a filter ensuring maximum compatibility with transfer functions relating to different HRTF head-related transfer function profiles

Step (8); Using a crosstalk filter with an angle greater than 3 degrees and less than 320 degrees

Step (9): Equalization of the listening medium in order to restore a signal identical to the original

Step (10): Specific processing at frequencies between 150hz and 300hz with binaural impulse responses including a 25% delay modification as well as a modification of the left signal of the right impulse response as well as the right signal of the left impulse response with a value between 10 and 20%.

Step (11): Specific processing for frequencies between 12000hz and 22000hz with specific processing for harmonics between 2 and 7 without modification of harmonic 1

Step (12): Treatment of frequencies below 150hz in mono via reduction of stereo tracks with application of an acoustic filter.

Step (13): Multiband 3D acoustic correction filter defined by the pickup distance from the reference system and the volume of the harmonics

Step (14): 3D Acoustic Correction filter applied to fundamental

Step (15): modification of the stereophonic image applied in step b) and to the fundamental of the signals corrected on the 3D signal

Step (16): Use of reference signals to generate the characteristic noise of vinyl

Step (17): Processing of the signal resulting from step (16) with a set of specific spatialized impulse responses.
Process for processing a digital sound signal according to Claim 1 or 2, characterized in that it comprises an additional step consisting in generating an imprint of a reference listening system in stereo and in binaural at a distance of between 20 and 300 cm and apply a filter to ensure maximum compatibility with HRTF profiles.
Process for processing a digital sound signal according to the preceding claim, characterized in that it comprises processing by a crosstalk filter having an angle greater than 3 degrees and less than 320 degrees.
Process for processing a digital sound signal according to at least one of the preceding claims, characterized in that it comprises an additional step consisting in applying an equalization processing to the listening medium in order to restore a signal identical to the original.
Process for processing a digital sound signal according to at least one of the preceding claims, characterized in that it comprises an additional step consisting in applying a specific processing to frequencies between 150hz and 300hz by impulse filtering with binaural impulse responses comprising a modification of delays of 25%, by application to each corrected signal of a low-pass filter and of weighted addition of said corrected signal and of the signal processed by impulse filtering.
Process for processing a digital sound signal according to the preceding claim, characterized in that said processing comprises a modification of a value of between 10 and 20% of the "left" component of the "right" channel of the signal processed by the impulse response as well than the "right" component of the "left" channel of the signal processed by the impulse response.
Process for processing a digital sound signal according to at least one of the preceding claims, characterized in that it comprises an additional step consisting in applying a specific processing to the frequencies comprised between 12000hz and 22000hz with a specific processing to the harmonics comprised between 2 and 7 without modification of harmonic 1, by application to each corrected signal of a high-pass filter and weighted addition of said corrected signal and of the processed signal.
Process for processing a digital sound signal according to at least one of the preceding claims, characterized in that it comprises an additional step consisting in applying a processing of frequencies below 150 Hz in mono via a reduction of the stereo tracks with application of a acoustic filter.
Process for processing a digital sound signal according to claim 1, characterized in that it comprises an additional step consisting in applying a multiband 3D acoustic correction filter defined by the pickup distance from the reference system and the volume of the harmonics.
Process for processing a digital sound signal according to at least one of the preceding claims, characterized in that it comprises an additional step consisting in applying a 3D acoustic correction filter applied to the fundamental.
Process for processing a digital sound signal according to at least one of the preceding claims, characterized in that it comprises an additional step of modifying the stereophonic image applied in step b) and to the fundamental of the signals corrected on the 3D signals.