T 0411/17 (Automatic adaptation to acoustic environments in a hearing aid/SONOVA) 19-05-2020

Reasons for the Decision

1. Patent in suit

The patent in suit aims at improving the automatic adaptation of a hearing aid to a changing acoustic environment. The hearing aid has an adjustable transfer function comprising one or more sub-functions to describe how audio signals received at the hearing aid's input relate to audio signals provided at the hearing aid's output. These sub-functions correspond to programs or algorithms that the hearing aid has at its disposal to cope with different acoustic environments. Each algorithm involves parameters to optimise its performance in view of the specifics of the acoustic environment prevailing at a particular time.

During the fitting of the hearing aid, a set of typical listening situations, i.e. a set of predetermined acoustic environments, may be considered and the hearing aid may be programmed with "predetermined base parameter sets" that define (the values of) the parameters required for the hearing aid's algorithms to cope best with each of those typical listening situations. The hearing aid is able to discriminate between listening situations by means of a classifier that analyses the current acoustic environment and determines how it relates to the predetermined acoustic environments.

For this purpose, the classifier regards the predetermined acoustic environments as "classes" and calculates, for each of those classes, a degree of similarity, i.e. a "class similarity factor", with the current acoustic environment. The class similarity factors are used to derive "activity parameter sets" from the predetermined base parameter sets, e.g. by weighting each base parameter set with a respective class weight factor derived from the corresponding class similarity factor. Such an activity parameter set prescribes, for its associated algorithm, the proper value of the parameters which the algorithm requires to accommodate the current acoustic environment. As a result, the hearing aid's transfer function can be adjusted to adapt to the current acoustic environment in an optimised way.

2. Main request: claim 1 as granted - novelty (Article 54(3) EPC)

2.1 Using the terminology of present claim 1, document D3 discloses

a) a method for operating a hearing device (paragraph [0001]) having an adjustable transfer function (paragraphs [0001] and [0030]: "Signalverarbeitung"; paragraphs [0002] and [0029]: "Weiterverarbeitung") comprising

M sub-functions ("Hörprogramm[en]"), wherein M is an integer with M >= 1 (paragraphs [0005], [0006] and [0013]), and wherein said transfer function describes how input audio signals generated by an input transducer unit of said hearing device relate to output audio signals to be fed to an output transducer unit of said hearing device (paragraphs [0002], [0003], [0005], [0006] and [0013]),

said method comprising the steps of

b) deriving said input audio signals from a current acoustic environment (paragraph [0002]; paragraph [0030]: "augenblicklichen Hörsituation");

c) for each of said M sub-functions (paragraph[0003], [0013] to [0015] and [0030]: the scheme of D3 is to be applied to all occurring listening situations):

d) deriving, on the basis of said input audio signals and for each class of N classes each of which describes a predetermined acoustic environment (paragraph [0003]: "Analyse des akustischen Eingangssignals zur Bestimmung der Hörsituation"; paragraph [0030]: "infolge einer Analyse der augenblicklichen Hörsituation"; given that, according to paragraph [0030], the parameters of the signal processing, and therefore the corresponding best fitting hearing-aid program, are set automatically based on an analysis of the momentary listening situation, each hearing-aid program must be mapped to a particular listening situation for which it performs best), a class similarity factor indicative of the similarity of said current acoustic environment with the predetermined acoustic environment described by the respective class (implicitly disclosed, otherwise an automatic setting of the hearing-aid parameters for the current listening situation as prescribed in paragraph [0030] would not be possible), wherein N is an integer with N >= 2 (paragraph [0003]: three common examples of typical listening situations are given);

e) deriving from N predetermined base parameter sets assigned to the respective sub-function (paragraph [0013]: each hearing-aid program is characterised by certain values for a set or subset of all possible parameters that may be relevant for a particular typical listening situation; by definition, the base parameter sets of the claim can each be chosen to encompass all those possible parameters) and in dependence of said class similarity factors (paragraph [0030]: "infolge einer Analyse der augenblicklichen Hörsituation") an activity parameter set (paragraph [0030]: "Dabei können die Signalverarbeitung betreffende Parameter auch automatisch durch die Steuereinheit 7 ermittelt und eingestellt werden, z.B. infolge einer Analyse der augenblicklichen Hörsituation" [emphasis by the board]: that the parameters can be determined based on the current listening situation directly and unambiguously means that they are derived, e.g. by means of an interpolation between two or more most relevant classes of typical hearing aid situations, starting from relevant base parameter sets corresponding to those classes that best fit the current listening situation) for the respective sub-function,

wherein each of said N base parameter sets assigned to the respective sub-function is assigned to a different class of said N classes (implicitly disclosed: a class corresponding to a typical listening situation will have attributed to it a hearing-aid program with a parameter set that best fits that class; a current listening situation may then be mapped directly onto one of those classes or to more than one via interpolation);

f) adjusting the respective sub-function by means of said activity parameter set (paragraph [0030]: "ermittelt und eingestellt" [emphasis by the board]); wherein for at least one of said M

sub-functions (paragraphs [0031] to [0033]: the "at least one" of the claim can be any of those hearing-aid programs which have the setting of the amplification, i.e. volume control, as a mandatory parameter), a time-averaged activity parameter set (paragraph [0033]: "Mittelungsalgorithmus"; once one parameter of the set is time-averaged, the entire parameter set can be considered as a

time-averaged set) is used for adjusting the respective at least one of said M sub-functions.

2.2 The appellant argued in view of Figure 2 of D3 that D3 did not anticipate feature f) of claim 1 because the averaging algorithm "Mittelungsalgorithmus" according to paragraph [0033] of D3 only applied to the starting values "Startwerte" and not to the activity parameter set. By contrast, the manual or automatic change of parameters such as the volume control as mentioned in paragraphs [0018] and [0030] was not time-averaged.

The board agrees with the appellant that the starting values in D3 may indeed not be active immediately after their determination and that they are stored in the memory for use when their associated hearing program is activated again. The board also accepts that the start values may be subsequently changed during operation, e.g. by the user manually adjusting the volume, as apparent from paragraphs [0018] and [0030] of D3.

However, as pointed out by the respondent, the wording of feature f) is broader than what is suggested by the appellant: it does not specify when the time averaging takes place and, in particular, it does not specify that the time-averaged activity parameter set is used instantaneously after its determination. As such, it encompasses a dynamic change of starting values during operation as well as a delayed use of updated starting values as determined via a time-averaging procedure. The claim allows, in particular, for a situation where the listening situation changes immediately after the time-averaging of the activity parameter set has been determined for a specific listening situation and where the result of the time-averaging is only used in the next instance at which this specific listening situation re-occurs. Moreover, as brought forward by the respondent, whether a particular hearing program is activated in D3 depends on the currently prevailing listening situation and on the classifier assessing that situation. Therefore, whenever the classifier in D3 concludes that a particular listening situation occurs, the corresponding hearing program is loaded into the processor with the starting value as determined via the time-averaging algorithm of paragraph [0033] of D3. Given that time averaging requires by definition a certain time span, the time averaging of paragraph [0033] of D3 must take place over a time period covering at least some of the previous occurrences when that particular listening situation prevailed. Consequently, the time-averaged starting value of paragraph [0033] anticipates feature f) of claim 1.

2.3 The appellant emphasised that D3 did not disclose a superposition ("Überlagerung") of at least two hearing programs in dependence of the current listening situation but, at most, taught an automatic selection of the best-fitting hearing program.

The board holds that claim 1 does not require such a superposition either. It merely requires to derive an activity parameter set from the predetermined base parameter sets in dependence of a class similarity factor. Consequently, the wording of claim 1 is broad also in this respect and a mere (automatic) selection of the best-fitting hearing program and its parameter set from the list of predetermined base parameter sets is, in itself, encompassed by this wording. Moreover, paragraph [0030] of D3 teaches to determine the parameters ("die Signalverarbeitung betreffende Parameter") and not, as alleged by the appellant, a selection of a program. Furthermore, in view of the wording "ermittelt und eingestellt" of paragraph [0030] of D3 as highlighted above, the disclosure of D3 goes beyond a mere selection and rather derives the activity parameter set from the predetermined base parameter set based on arithmetic operations.

2.4 The appellant argued further that, in view of the fact that claim 1 requires at least two classes but only at least one sub-function, the hearing programs of D3 constituted the N classes rather than the sub-functions of the claim.

In that regard, the board notes that the skilled reader would, however, readily understand from the patent in suit, in particular from granted claim 1, that the number of classes and sub-functions in the claim is not indicative of any difference between them: one

sub-function can correspond to multiple classes (in the sense of listening situations), namely by using different parameter sets. Moreover, as apparent from paragraphs [0026], [0030], [0050] and [0051] of the patent in suit, the classes of the claim clearly correspond to the "predetermined acoustic environments" and not to "hearing aid programs". Consequently, the hearing programs of D3 correspond to the "sub-functions" and the listening situations of D3 correspond to the "classes" of the claim. Moreover, these classes must correspond to the listening situations as referred to by the term "Hörsituation[en]" in paragraph [0003] and the term "Hörumgebungen" of paragraphs [0011], [0015] and [0030] of D3.

By contrast, paragraph [0035] of the patent in suit describes the sub-functions as "meaningfully combined parts" of the hearing aid's transfer function through which many types of signal processing could be realised. This would mean, for a skilled reader, that the sub-functions are to be seen as algorithms that provide for a specific function within the signal processing scheme of the hearing aid. Furthermore, as set out in paragraphs [0035], [0037] and [0040] of the patent in suit, the sub-functions are controlled by the activity parameter sets which are dynamically adapted to the current acoustic environment. This is equivalent to the automatic adaptation by processor 7, induced by an analysis of the current acoustic environment, of the parameters relevant to the signal processing in processor 2 of paragraph [0030] and the hearing program of paragraph [0032] of D3. Therefore, the signal processing scheme according to paragraph [0030] of D3, as implemented via the hearing program of paragraph [0032] of D3, corresponds to the sub-functions of the claim and of the aforementioned paragraphs [0035], [0037] and [0040] of the patent in suit.

2.5 Hence, D3 discloses all the limiting features of claim 1 of the main request. Consequently, the subject-matter of present claim 1 is not new under Article 54(3) EPC.

3. First auxiliary request: claim 1 - novelty (Article 54(3) EPC)

3.1 The first auxiliary request is identical to the first auxiliary request underlying the decision under appeal. Claim 1 of this request differs from the one of the main request essentially in that it further specifies that (board's underlining)

g) the activity parameter set is obtained as a mixture of base parameter sets, the mixture depending on the class similarity factors of the base parameter sets.

3.2 The board notes that the expression "augenblickliche[n] Hörsituation" in paragraph [0030] of D3 refers to the current listening situation, which may differ from the set of typical listening situations, i.e. the "classes" of claim 1, that are stored in the hearing aid and that are referred to in e.g. paragraph [0003] of D3. When such a difference occurs, the skilled person knows that control unit 7 of paragraph [0030] is configured to choose one of two options: either (i) it selects that particular typical listening situation that best fits the current listening situation, which directly determines the parameter set to be used, or (ii) it makes an interpolation of the best-fitting typical listening situations and, correspondingly, an interpolation of the parameter sets attributed to those typical listening situations. The former approach requires less calculations whereas the latter approach normally allows for a better reproduction quality.

D3 does not explicitly state which approach is followed. However, given that there are only two approaches to choose from, there is a direct and unambiguous indication by the expressions "augenblicklichen Hörsituation" and "ermittelt" in paragraph [0030] to determine the currently active settings, i.e. the activity parameter set, as a mixture of base parameter sets. The board notes again the use of the words "ermittelt[en]" and "Ermittlung" in paragraphs [0014] to [0016] and [0032] of D3, which suggest that these terms are used in D3 within the context of calculations (see also the word "Berechnung" in paragraph [0032]).

Given that only the latter of the two approaches mentioned above requires a "calculation" of the parameter set (the former one deriving the parameter set directly from the selected typical listening situation), D3 in fact implies an application of the latter approach.

3.3 The appellant argued that paragraphs [0014] to [0016] and [0032] of D3 could not be used to interpret paragraph [0030], because the former determined the start value of the hearing aid's parameter whereas the latter concerned the change of the parameter value during the hearing aid's operation.

However, the board's reference to paragraphs [0014] to [0016] and [0032] of D3 in point 3.2 above only aims at how the expressions "ermittelt[en]" and "Ermittlung" used in paragraph [0030] are to be construed. For this purpose, it is immaterial that paragraphs [0014] to [0016] and [0032] and paragraph [0030] concern a possibly different value of the parameters. In addition, the board's argumentation that the interpolation of the parameter sets constitutes a direct choice for the skilled person also holds without the reference to these paragraphs. It follows from the above that D3 likewise anticipates feature g) of present claim 1.

3.4 Hence, D3 discloses all the limiting features of claim 1 of the first auxiliary request. Consequently, the subject-matter of present claim 1 is not new either (Article 54(3) EPC).

4. Second auxiliary request

4.1 Claim 1 of the second auxiliary request differs from that of the first auxiliary request in that it additionally includes the following methods steps (board's underlining):

h) choosing an averaging time for said time-averaging in dependence of past changes in the respective activity parameter set;

i) decreasing said averaging time when said past changes in the respective activity parameter set decrease;

j) increasing said averaging time when said past changes in the respective activity parameter set increase.

4.2 Admission into the appeal proceedings (Article 12(4) RPBA 2007)

4.2.1 The second auxiliary request is identical to the second auxiliary request underlying the decision under appeal. The opposition division did not admit that request into the opposition proceedings on the grounds that it was late-filed (filed during the first-instance oral proceedings) and that claim 1 of the request was not based on the claims as granted.

4.2.2 Article 12(4) RPBA 2007 conveys the board the discretionary power to hold inadmissible facts, evidence or requests which were not admitted in the first-instance proceedings.

4.2.3 The respondent requested to confirm the opposition division's assessment and not to admit the second auxiliary request into the appeal proceedings. The respondent emphasised that the patent proprietor could have foreseen from the annex to the summons of the opposition division that the opposition division would raise an objection under Article 54(3) EPC against the main request in view of D3 and an objection under Article 54(2) EPC against the auxiliary request in view of D9 as set out in the decision under appeal. In the respondent's view, the patent proprietor deliberately chose not to file the second auxiliary request before expiry of the time limit pursuant to Rule 116(2) EPC.

4.2.4 The board holds that the opposition division had a discretion not to admit the second auxiliary request into the proceedings. Moreover, a board may overrule the way in which the department of first instance had exercised its discretion if it came to the conclusion either that it had not done so in accordance with the proper principles or had done so in an unreasonable way, and had thus exceeded the proper limits of its discretion. It is apparent from the file that the opposition division exercised its discretion not to admit the request into the proceedings on the grounds that, in view of feature g), claim 1 of the second auxiliary request did not find a correspondence in a dependent claim of the patent as granted. It considered the subject-matter of the amendments to be technically complex, such that the parties could not be expected to deal with it at oral proceedings (see appealed decision, Reasons, point 3).

4.2.5 The board agrees with the appellant that feature g) had been introduced into the independent claims with the first auxiliary request filed with a telefax of 25 September 2015, i.e. more than a year ahead of the oral proceedings of 30 November 2016 before the opposition division. It is also apparent to the board that the subject-matter of claim 1 of the second auxiliary request corresponds to a combination of claims 1, 8 and 9 of the then first auxiliary request. The board therefore considers that there was sufficient time for the opponent to prepare for the subject-matter of the second auxiliary request, the more so since the opponent could have foreseen that auxiliary requests were to follow given the preliminary opinion of the opposition division as set out in the annex to its summons to oral proceedings. The board notes also that, apart from feature g), claim 1 of the second auxiliary request corresponds to a combination of claims 1, 8 and 9 as granted. At no point in time did the opponent raise objections against the subject-matter of such a combination, although this could have been done as early as with the notice of opposition.

4.2.6 In view of the above, the board, exercising its own discretion under Article 12(4) RPBA 2007, decided to admit the second auxiliary request into the appeal proceedings.

4.3 Novelty (Article 54 EPC)

4.3.1 Claim 1 of the second auxiliary request is novel over D3 in view of the wording of feature g). D3 only shows a time averaging by the "Mittelungsalgorithmus" according to paragraph [0033], but the specifics of feature g) are not disclosed therein.

4.3.2 Present claim 1 is also novel over document D9 and document D1 in view of at least feature g).

4.3.3 Novelty of present claim 1 with respect to D3, D9 and D1 was not contested by the respondent at the oral proceedings before the board.

4.3.4 In view of the above, novelty of claim 1 of the second auxiliary request is acknowledged (Article 54(2) and 54(3) EPC).

4.4 In sum, the present second auxiliary request is admitted into the appeal proceedings and is held to be allowable under Article 54 EPC.

5. Remittal of the case to the opposition division (Article 111(1) EPC; Article 11 RPBA 2020)

5.1 Pursuant to Article 11 RPBA 2020, the board shall not remit a case to the department whose decision was appealed for further prosecution, unless special reasons present themselves for doing so.

5.2 The opposition division had to decide only on the matters of sufficiency of disclosure (Article 83 EPC) and novelty (Article 54(2) and (3) EPC) as regards the present main request and first auxiliary request. As a consequence, an assessment of inventive step in view of the available prior-art documents was not necessary.

Given that several lines of argument concerning lack of inventive step were submitted by the opponent in the notice of opposition, a ruling on those attacks by the board for the very first time in these appeal proceedings could not be given without an undue burden and would run contrary to the very purpose of a judicial review within the meaning of Article 12(2) RPBA 2020. Moreover, this course of action was explicitly consented to by both parties at the oral proceedings before the board. The board considers that the above observations represent "special reasons" within the meaning of Article 11 RPBA 2020 for remittal of the case.

5.3 Consequently, the board has decided to set the decision under appeal aside and to remit the case to the opposition division for further prosecution, in particular for the assessment of inventive step, on the basis of claims 1 to 14 of the second auxiliary request on file, i.e. of the previous "sixth auxiliary request" filed with the letter dated 27 April 2017.

6. Substantial procedural violation - request for reimbursement of the appeal fee

6.1 According to Rule 103(1)(a) EPC, the appeal fee shall only be reimbursed in the event that the board deems the appeal to be allowable.

6.2 Since the decision under appeal is to be set aside, the appeal is indeed deemed to be allowable. The board, however, considers that no substantial procedural violation occurred in the opposition proceedings. The opposition division applied the proper principles to exercise its discretion not to admit the second auxiliary request into the proceedings on the basis of the correct facts (i.e. late-filed request; including features from the description) and in a reasonable way (i.e. by giving conclusive grounds for their discretionary decision). This does not constitute a procedural violation, let alone a substantial one, according to the established jurisprudence of the Boards of Appeal. The fact that the board used its discretion in a different way does not militate against the above conclusion (see e.g. G 7/93, Reasons, point 2.6).

6.3 Therefore, the board sees no reason to order a reimbursement of the appeal fee.