From Interval Recognition to Absolute Pitch: A Dialogue on Inner Hearing, Pitch Memory and a Training Path

Date: 20 October 2026

Editorial note: This record has been edited from the day’s conversation, following the original order of discussion. Speech recognition errors, typos, slips of the tongue and clearly awkward phrasing have been corrected, and some repetitive questions and explanations have been removed. The development and clarification of ideas, disagreements and revisions have been retained. This is an edited record of a conversation, rather than a verbatim transcription of an audio recording. Explanations of cognitive mechanisms presented here are judgements or hypotheses discussed in the conversation, not conclusions established by experimental evidence.

User: Why do we have perfect unisons, perfect fourths and perfect fifths, while the other interval numbers are not named in the same way?

Assistant: Western music theory groups perfect unisons, perfect fourths, perfect fifths and perfect octaves as “perfect intervals”. This classification has historical and acoustic roots. In just intonation, the unison, octave, fifth and fourth correspond to simple frequency ratios of 1:1, 2:1, 3:2 and 4:3 respectively, and they were given a special status as consonances in traditional music theory. Seconds, thirds, sixths and sevenths instead use “major” and “minor” as their basic classifications. However, “perfect” is a historically established theoretical term; it does not mean that other intervals are impure or unpleasant. Nor do the frequency relationships in modern twelve-tone equal temperament exactly match all these simple integer ratios.

User: So these names are largely a matter of convention.

Assistant: That is a reasonable way to understand them. The names are largely historical conventions, but they are not entirely arbitrary: they also have a basis in acoustics and traditional musical practice. The acoustic relationships between intervals are objective, while their classification and naming involve historical conventions. A different set of names would not change the actual relationships between the sounds.

User: When training our ears, then, should we remember these names, or remember—or recognise—how many semitones apart the notes are?

Assistant: The central aim of ear training should be to recognise the relationship between two sounds directly. Interval names and semitone counts are both useful, but they serve as tools for understanding and checking. A minor third spans three semitones, a major third four, a perfect fourth five and a perfect fifth seven. Ideally, you hear two notes and recognise their relationship directly, rather than first counting semitones and then working out the name. One further point is that the same number of semitones does not necessarily correspond to the same theoretical name: an augmented fourth and a diminished fifth both span six semitones.

User: But practising in C for now will not prevent me from recognising interval distances in other keys later. I can simply change the key, rather like transposing. For example, if I know a scale very well in this key, I can immediately recognise the same kind of scale in another key.

Assistant: Your point makes sense. Intervals remain the same under transposition: C–E and D–F♯, for example, are both major thirds. Establishing an aural reference in one key can indeed be an effective training method. However, familiarity with a scale in one key is not quite the same ability as recognising the interval between any two notes without a fixed tonal context. You can first become secure in C, then use other starting notes to check whether the skill transfers. Keeping a reference fixed and varying it are compatible approaches.

User: I can now distinguish most of the intervals from a perfect unison up to around a perfect fifth or a sixth, including the intervening semitone steps.

Assistant: If you can also distinguish interval sizes that differ by a semitone within that range, your interval discrimination is becoming quite fine. Another useful distinction is between hearing that two intervals differ and immediately identifying a particular interval when you hear it. These represent different degrees of fluency. Do you hear the interval distance directly, or do you need to count semitones in your head?

User: I hear the interval distance directly. I am not counting semitones.

Assistant: Then your judgement comes mainly from direct auditory perception, without relying on conscious semitone calculation. That is precisely the ability interval training aims to develop. More specifically, a person might recognise an interval through a sense of pitch distance, or through the distinctive character of its overall sound. Both may coexist in the listening experience.

User: What is the difference between the distinctive sound character you mentioned and the sense of distance?

Assistant: By a “sense of distance”, I mean that when you hear C–D, C–E and C–G, the second note seems progressively further from the first. By “sound character”, I mean the different overall impressions you have of a major third, a minor third, a perfect fifth and so on, including their sense of blending, consonance and characteristics acquired through long familiarity. There is no absolute boundary between these experiences, and a subjective description alone cannot establish which mechanism the brain is actually using.

User: I think I can certainly tell when the size of an interval changes. I can also hear the differences in their musical effects.

Assistant: It seems, then, that your perception includes both interval distance and musical effect. The size of an interval is fixed, while its harmonic function and expressive effect in a particular piece also depend on the key and musical context. Your training goal now could be to make this judgement more reliable and to develop the ability to imagine the corresponding intervals actively without hearing them played.

User: Yes. At the moment, these semitone steps are difficult both to sing and to form as corresponding music in my mind.

Assistant: There are three related but distinct processes here: recognising an interval after hearing it; actively forming a pitch or interval internally without an external sound; and accurately singing the sound you imagine. Your description at that point seemed to suggest that recognition was already fairly fluent, while inner hearing and vocal reproduction still needed development. In particular, actively imagining notes with sharps or flats can be practised independently; singing them need not be the test.

User: What I just said was not entirely accurate. There has actually been a noticeable change over the past day or two. I can not only clearly distinguish intervals that differ by a semitone, but am also gradually becoming able to form sounds separated by several semitones actively in my mind. After repeated listening, a kind of inaudible sound echoes in my head, and the pitch of this inner resonance is now fairly accurate. Singing it may take a little more time, but that is no longer important, because I am not training as a singer or working on vocal technique.

Assistant: Then I need to revise my earlier judgement. Your description of an “inaudible sound echoing” is close to what music psychology calls auditory imagery. What matters to you now is forming musical sounds with definite pitch relationships in your mind, without actually vocalising them. Singing accurately also involves vocal motor control, so it should not be the main measure of this training. To check the accuracy of an imagined pitch, you can occasionally imagine it first and then compare it with an actual played note.

User: I do not need to hear it first. With a C and an F, for example, those sounds appear directly in my mind. It is not a short-term memory left by something I have just heard.

Assistant: That distinction matters. You are describing actively recalling sounds from long-term memory, rather than simply retaining a sound briefly after hearing it. This can be called voluntary auditory imagery, and it does not necessarily require a recently played sound as a cue. We should still distinguish between imagining the C–F interval accurately and imagining the fixed absolute pitches of C and F themselves. These may coincide, but they should be assessed as separate abilities. For instance, can you directly raise the F you have just imagined by a semitone, turning it into F♯?

User: Yes, I can.

Assistant: Then the ability you describe includes both actively recalling an interval and deliberately changing a note’s pitch in your mind. That is very useful for piano playing, reading music and musical invention. Still, feeling that the semitone change is accurate and being objectively accurate every time are different matters; comparison with actual played sounds is needed to check this.

User: Although I have only used this software for training over the past two or three days, I established an internal musical reference many years ago: C major and the C major triad. I remember C, E and G—Do, Mi and Sol—very accurately, almost with the precision of absolute pitch. So I already had that foundation. With those three pitches in place, it is relatively easy to remember, identify and actively produce the others.

Assistant: That explains why you are not starting from scratch. Your longstanding pitch memories for C, E and G very likely provide a stable reference for your recent listening practice. We can distinguish long-term pitch memory from relative pitch: the former retains particular pitches, while the latter handles relationships between notes. Your recent software practice may be refining and extending auditory representations you already had, rather than creating a new ability from nothing in just a few days.

User: That is exactly right. Even across several octaves, I can play C in different registers without looking, and the sound matches what I had already formed in my inner hearing.

Assistant: That makes your account of long-term absolute pitch memory for C more convincing. You are describing more than familiarity with the name C: you can actively form an expectation of its pitch in different registers without an external sound. The scientific distinction to retain is that highly accurate long-term memories for a few pitches do not automatically demonstrate stable absolute pitch across all twelve pitch classes. They do, however, provide a foundation well worth testing further.

User: When I play a complex piece, if one note is wrong, I hear it immediately and very clearly. It is not just that my fingers feel a mistake; the sound itself seems strikingly wrong.

Assistant: That suggests you have probably formed clear internal expectations of the melody, harmony and movement of the voices as you play. If the sound your fingers produce differs from that expectation, you notice immediately. This relates to auditory prediction in music cognition. However, detecting a wrong note does not necessarily depend entirely on absolute pitch: good relative pitch and familiarity with the piece can also contribute. Once you notice the wrong note, can you immediately hear the correct one internally without looking at the score?

User: Of course. That comes very naturally to me. It is the same when I listen to other people play, not just when I play myself. If a note is wrong, I recognise it immediately. Even in a very complex, densely textured passage, a single wrong note stands out clearly to me.

Assistant: Then the ability cannot simply be attributed to feedback from your finger movements. When listening to someone else, you can also detect a deviation through hearing and an existing musical expectation. However, recognising a departure from a familiar piece differs from deciding that a note is “objectively wrong” in a completely unfamiliar work without a score. In an unfamiliar piece, a note that seems inconsistent with common tonal or harmonic patterns does not necessarily mean the performer has made a mistake. Your experience at least shows a strong sensitivity to small deviations in familiar music.

User: I actually think absolute pitch can be trained. By establishing a reference pitch internally over a long period, you can gradually build a frame of reference for absolute pitch.

Assistant: I think that is a training hypothesis worth taking seriously. Repeatedly memorising a fixed pitch such as middle C at first may help establish a fairly stable internal anchor, which can then be extended to more pitches. Existing music cognition research also supports the possibility that some adults can improve absolute pitch identification through training, although the accuracy, range and stability they eventually achieve vary between individuals.

Three studies were cited in the discussion. A small study published in PLOS ONE in 2019 found that, after eight weeks of training, some of six adult participants with good auditory memory performed at levels approaching those of traditional absolute pitch possessors. Across a series of experiments published in Attention, Perception, & Psychophysics in 2020, six of 43 participants reached the study’s criterion of at least 90% identification accuracy across twelve pitch classes. In a 2025 study in Psychonomic Bulletin & Review, twelve adult music learners, after eight weeks of training, could identify an average of about seven pitch classes with at least 90% accuracy in the training task. These findings support the trainability of absolute pitch identification in adulthood, but do not establish that everyone can reach the same level.

I also raised a distinction: deriving other notes from C is not the same process as directly recognising the absolute pitch of each note.

User: The absolute pitch ability I mean does not involve starting from a reference C and then identifying another note through an interval relationship. I recognise it directly, just as I recognise notes on a staff. I do not count them one by one. Each pitch is remembered individually.

Assistant: That corrects my earlier understanding of your point. You mean that each pitch can acquire its own auditory representation in long-term memory, to be recognised or recalled directly when needed, without C having to serve as an intermediate step in conscious awareness. The staff-reading analogy is apt: a beginner may count the lines, but a fluent reader recognises a note immediately. Visual symbol recognition and auditory pitch recognition do not have identical mechanisms, but both may involve a learning process that moves from conscious reference to automatic recognition.

User: That was how it began, of course. Many years ago, I did first strengthen my memory of middle C and establish an anchor. Later, I could move beyond that anchor.

Assistant: That gives a coherent account of the learning process over time: first establishing a stable reference through middle C, then gradually forming long-term memories for other pitches, until you no longer need to call on the anchor consciously when recognising them. From a cognitive science perspective, however, subjective experience alone cannot determine whether the brain has entirely stopped using unconscious reference processing. What you are proposing is a path from learning with an anchor to direct recognition.

User: Whether the brain is secretly using that anchor automatically and rapidly no longer really matters. At least in terms of outward performance, the ability is already that of absolute pitch.

Assistant: I agree that functional performance and its internal implementation should be distinguished. Absolute pitch identification should mainly be judged by whether someone can identify pitches quickly, accurately and repeatedly without an external reference, rather than by requiring proof that no implicit reference process occurs in the brain. However, confirming complete and stable absolute pitch across all twelve pitch classes also requires examining accuracy, response time and performance across different registers and timbres.

User: This training path may therefore be a viable way of developing absolute pitch that I happened to discover without intending to. For me personally, its viability comes from direct experience. Of course, the ability still needs to be strengthened. It is currently somewhat unstable, or rather, fairly stable direct recognition has developed for only a few notes. My responses to other notes may be slower, but a slower response does not mean I am deriving or calculating an interval. It is still direct recognition; memory retrieval is simply not yet as quick. So the next step is to strengthen each pitch individually and gradually establish this ability across all the notes.

Assistant: You have now described both the learning process and your current position clearly. The key distinction to retain is that some notes already have stable, rapidly accessible individual memories, while others are less familiar and take longer to recognise directly. A slower response does not necessarily mean conscious interval calculation. It may also relate to familiarity, retrieval speed and the stability of long-term memory, although the specific internal mechanism still needs further investigation.

The next stage could focus on each of the twelve chromatic pitches. First, listen repeatedly to a note, actively recall it without an external sound, and then check it on the piano. Follow this with random single-note identification, recording accuracy and response time for each pitch separately. Once performance is stable in a fixed register, vary the octave and the instrument’s timbre. The aim is to develop auditory representations of each pitch that are as stable and directly accessible as possible, rather than forcing yourself to calculate quickly from an anchor.

Your personal experience supports this as a training path grounded in practice and worth continuing to explore and test. However, evidence that it helps you does not establish that it will work equally well for everyone, nor does it establish the mechanism by which complete absolute pitch develops. Your clearest training direction now is to build on the long-term pitch memories you already have for C, E and G, and further consolidate independent recognition and inner reproduction of the other pitches.

Appendix: Research referred to in the conversation (checked when archived)

Van Hedger, Heald & Nusbaum (2019), “Absolute pitch can be learned by some adults”, PLOS ONE.
https://doi.org/10.1371/journal.pone.0223047

“Is it impossible to acquire absolute pitch in adulthood?” (2020), Attention, Perception, & Psychophysics.
https://doi.org/10.3758/s13414-019-01869-3

“Learning fast and accurate absolute pitch judgment in adulthood” (2025), Psychonomic Bulletin & Review.
https://doi.org/10.3758/s13423-024-02620-2

— End of edited conversation —