A convolutional neural network that listens to piano audio and transcribes it into a MIDI piano roll — predicting which of the 88 piano keys are active at each moment in time.
This project trains a CNN to perform automatic piano music transcription: given a raw audio recording of a piano piece, the model outputs a binary piano roll indicating which notes are sounding at each time frame. Audio is first converted to a Constant-Q Transform (CQT) spectrogram with 264 frequency bins (3 bins per semitone across all 88 piano notes), which aligns frequency resolution directly to the piano keyboard. The CNN processes this representation through three convolutional layers and a frequency-pooling step that maps the 264 CQT bins down to 88 note bins, then applies a 1×1 output convolution to produce per-note activation logits at every time step. The model is trained on the MAESTRO v3.0.0 dataset — a large collection of professional piano recordings with precisely aligned MIDI ground truth — using binary cross-entropy loss with positive-class weighting to handle the sparsity of active notes (~2–3% of note-time slots at any given moment).
# 1. Install dependencies
pip install -r requirements.txt
# 2. Place the MAESTRO v3.0.0 dataset folder in the data folder
# Expected path: data/maestro-v3.0.0/
# 3. Run the overfitting smoke test (sanity check on 10 samples)
python train.py # set MODE = "overfit" in __main__
# 4. Visualize a CQT spectrogram + piano roll pair (input + ground truth output)
python train.py # set MODE = "visualize" in __main__
# 5. Run full training
python train.py # set MODE = "train" in __main__See SETUP.md for full installation instructions and dataset download steps.
| Video | Link |
|---|---|
| Demo Walkthrough | [https://youtu.be/fFMZ1TUVFkg] |
| Technical Explanation | [https://www.youtube.com/watch?v=AmWqqpGAl_E] |
See Evaluation.md/Evaluation.pdf (depending on how you would prefer to render) for more information regarding specific evaluation iterations and model improvement. Summary is provided below:
The model is evaluated on frame-level note detection using precision and recall at a decision threshold of 0.5. Loss alone was not an effective metric due to it not being able to diffferentiate effectively between precision or recall failures, and some low loss strategies include just randomly guessing high probability notes.
Used to verify the model can learn before scaling to full training:
Summary Matrix:
| Precision | Recall | Loss | Probs mean | |
|---|---|---|---|---|
| BCE (0.5) | 0.29 | 0.559 | 0.2655 | 0.136 |
| BCE (0.6) | 0.355 | 0.350 | 0.278 | 0.147 |
| Focal (0.5) | 0.315 | 0.508 | 0.073 | 0.156 |
| Focal (0.6) | 0.312 | 0.138 | 0.0948 | 0.134 |
Key observations:
- The CQT representation with 3 bins/semitone gave noticeably better frequency resolution than mel spectrograms or raw STFT for distinguishing adjacent piano notes.
- Recall improves faster than precision due to the sparse nature of piano rolls; positive-class weighting (pw=5) balances this tradeoff.
- The model achieves decreasing loss across all 20 overfit epochs, confirming the architecture can learn from the data.
| Feature Representation | Notes | Behavior |
|---|---|---|
| Mel (128 bins) | Insufficient freq resolution | Adjacent notes collapse to same bin |
| STFT (2049 bins) | Linear frequency scale | High resolution but poorly suited to pitch |
| CQT (264 bins) | Log-frequency, 3 bins/semitone | Aligns with piano keyboard — best for transcription |
Detailed Evaluation and Analysis (appended from Evaluation.md/pdf, see PDF for higher fidelity images):
- Audio converter process (Spect -> STFT -> CQT)
- Mel Spectrogram:
- The first iteration audio processor used a Mel Spectrogram with a 512-sample FFT window and 128 Mel filter banks. The Mel scale seemed like a natural starting point since it directly mirrors how humans hear pitch, which seemed to be a reasonable way to help a model learn to interpret it the same length.
- This is what the Mel Output looked like on top, with the corresponding piano keys being mapped out below. From here, it looks like it preserves a lot of the pitch of the higher keys, however, the scale from the spectrogram to the piano roll is confusing (128 → 88), and therefore makes it much harder for the model to map specific Mel banks to specific notes. For instance, the lowest notes create an extreme amount of frequency bleed across basically all Mel filter banks, which is not easy for a model to interpret. The output would therefore read the attacks of initial note presses well, but it would have no frequency context and therefore predict the entire After aggressively tuning the model to try to pull more information from this representation, research indicated that it would be more effective to use a higher fidelity representation to get enough data for our model.
- STFT: second iteration, provided more frequency context, (output had 2049 channels, much higher fidelity, but scaling was linear as opposed to log)
- The second iteration used a raw STFT output, using the same FFT window of 512. This produced 2049 frequency channels, which gave our model so much more frequency information to work with.
- The detection improved marginally, and there was a lot less aggressive frequency bleed. This constrained the probabilities to specific notes, however, the STFT’s linear scaling made it a lot harder for the model to generate note distinction, so it ended up instead predicting the same note quite often. This is because most of the dataset was in the lower registers, where fewer bins covered more notes, so there was less fidelity at that part of the STFT which made it harder for the model to differentiate those parts. Even visually, in the plot above, while the harmonics rise far above into the rest of the plot, the highest power elements are all clustered at the bottom of the chart, which shows that the fidelity is in the wrong spots of the graph, so to speak.
- CQT: Constant Q transform: Final iteration, basically a Log scaled equivalent of the STFT, but able to create multiple bins per octave, get better note visibility
- These iterations led us to the CQT (constant Q transform). This was a happy medium between the two previous iterations, where it creates high fidelity frequency output, but its frequency bins are spaced logarithmically. This means that every octave gets equal representation (piano note frequencies are logarithmic, not linear).
- As visualized in the transformation above, the CQT gives a very clear image of the highest power notes across the whole frequency spectrum. We still see less clarity at the lower frequencies, but this is now mostly due to the articulation of the performers emphasizing the melody (higher notes), making the lower notes less interpretable. From here, we have an accurate baseline that our model can use to pick out the notes much more accurately.
- The progression from Mel → STFT → CQT was driven entirely by evaluation, both quantitatively from precision and accuracy scores, as well as visually observing what the prediction map looked like at the output. Each stage identified a specific failure mode (insufficient bin depth, then linear frequency mismatch), and each architectural change directly addressed that failure while the transcription metrics confirmed the improvement.
- Mel Spectrogram:
- Positive weighting/regularization/bias (USING BCE)
- Positive weighting (pw) varied to get the output to be properly balanced, as opposed to predicting far too many notes or far too few notes. At first, the code had a big issue with not nearly enough note probabilities rising above the positive threshold, so using a low pw (3) helped reduce that initial sparsity. However, that resulted in pure randomness and about half the notes on at any given time. All specific stats for all pw inputs are appended below, but importantly, we can see a very high recall and low precision, which indicates that there are many false positives in this example. Additionally, the predicted probabilities mean was around 0.5, whereas the mean of our target was 0.027, so the probabilities were skewed way too high.
- pw=3:
- Epoch 19, Loss: 0.7800253033638
- logits mean/std: 0.14166559278964996 0.47083476185798645
- probs min/mean/max: 0.17786651849746704 0.5309885144233704 0.9780147075653076
- target mean: 0.027157314121723175
- precision: 0.045961346477270126 recall: 0.970251739025116
- pw=3:
- Therefore, increasing the pw to 5 allowed for the precision to increase marginally and was the first way that the model was able to move away from being centered at 0.5. However, as visualised in the data below, it was not super effective in reducing this. The mean only shifted by ~0.01, with marginal precision and recall improvements.
- pw=5:
- Epoch 19, Loss: 0.6161904335021973
- logits mean/std: -0.2741478681564331 0.5397118926048279
- probs min/mean/max: 0.1318133920431137 0.4318477213382721 0.9563058614730835
- target mean: 0.027157314121723175
- precision: 0.11925160884857178 recall: 0.8850114345550537
- pw=5:
- This did not significantly improve anything. Therefore, the next step was introducing an overall bias into the model. The pw modification did not significantly change the mean of the output, and since the mean output (0.43) is far greater than the target mean of 0.027 that the model is trying to optimize for. Therefore, the best way to rectify this is to bias the output layer so the sigmoid function doesn’t split around 0.5, but rather around 0.027, where the actual mean should be. Therefore, by adding this bias term before any training commences, the model output improves significantly.
- model.head.bias.data.fill_(-2)
- Epoch 19, Loss: 0.2656612694263458
- logits mean/std: -2.219456434249878 1.0583772659301758
- probs min/mean/max: 0.004196124617010355 0.1363288164138794 0.9547777771949768
- target mean: 0.027157314121723175
- precision: 0.29081329703330994 recall: 0.559333086013794
- model.head.bias.data.fill_(-2)
- The model had an extremely better performance. The target mean was much closer to the target threshold (0.136), which while overpredicting to some extent, did a fantastic job of balancing a reasonable recall with a much higher precision than any other example.
- Positive weighting (pw) varied to get the output to be properly balanced, as opposed to predicting far too many notes or far too few notes. At first, the code had a big issue with not nearly enough note probabilities rising above the positive threshold, so using a low pw (3) helped reduce that initial sparsity. However, that resulted in pure randomness and about half the notes on at any given time. All specific stats for all pw inputs are appended below, but importantly, we can see a very high recall and low precision, which indicates that there are many false positives in this example. Additionally, the predicted probabilities mean was around 0.5, whereas the mean of our target was 0.027, so the probabilities were skewed way too high.
- Loss Function: BCE vs. Focal
- BCE Loss: Using BCE loss, able to get here: (pw =5, head.bias = -2) (cutoff 0.5)
- ![][image4]
- Final Stats:
- Epoch 19, Loss: 0.2656612694263458
- logits mean/std: -2.219456434249878 1.0583772659301758
- probs min/mean/max: 0.004196124617010355 0.1363288164138794 0.9547777771949768
- target mean: 0.027157314121723175
- precision: 0.29081329703330994 recall: 0.559333086013794
- This was a really good place to start, the model recognized the melody pretty well, but had issues thresholding out some of the higher frequencies. However, Focal Loss is supposed to be optimized for imbalanced datasets like the piano performances this model is training on, so it possibly would be much better at dealing with these imbalances than plain BCE. An alternative would be to increase the probability threshold, so experimenting with both can illustrate the best possible course of action.
- BCE Loss: (cutoff 0.6)
- ![][image5]
- Final stats:
- Epoch 19, Loss: 0.27767711877822876
- logits mean/std: -2.0682196617126465 0.9819927215576172
- probs min/mean/max: 0.01243616547435522 0.14710761606693268 0.9497251510620117
- target mean: 0.027157314121723175
- precision: 0.3557063937187195 recall: 0.34692710638046265
- All this does is reduce the presence of the low frequency notes,
- Focal Loss: (cutoff 0.5)
- ![][image6]
- Final stats:
- Epoch 19, Loss: 0.07325195521116257
- logits mean/std: -2.0537948608398438 1.1131919622421265
- probs min/mean/max: 0.00642482889816165 0.1577495038509369 0.9094279408454895
- target mean: 0.027157314121723175
- precision: 0.3147444725036621 recall: 0.5078456997871399
- Looking at this output is really interesting, because at first glance, it seems that it is a little bit of a simpler output than the BCE at 0.5 threshold. However, the numbers tell a different story. While this output is slightly higher precision and significantly lower loss, the recall is much further off, and the probability mean is much further off. Additionally, upon a closer examination of the Focal loss inference, a lot of the detail/overarching pattern of the notes are lost.
- Focal Loss: (threshold 0.6)
- ![][image7]
- Final Stats:
- Epoch 19, Loss: 0.09484720230102539
- logits mean/std: -2.3492205142974854 1.2376179695129395
- probs min/mean/max: 0.0036973310634493828 0.1337791085243225 0.8576645851135254
- target mean: 0.027157314121723175
- precision: 0.3124304413795471 recall: 0.13762667775154114
- With a higher threshold, this impact is even more apparent. Barely any detail remains, and it's very hard to tell what the actual song is doing. This threshold is far too high and therefore needs to be reduced in order to be effective.
- BCE Loss: Using BCE loss, able to get here: (pw =5, head.bias = -2) (cutoff 0.5)
Summary Matrix:
| Precision | Recall | Loss | Probs mean | |
|---|---|---|---|---|
| BCE (0.5) | 0.29 | 0.559 | 0.2655 | 0.136 |
| BCE (0.6) | 0.355 | 0.350 | 0.278 | 0.147 |
| Focal (0.5) | 0.315 | 0.508 | 0.073 | 0.156 |
| Focal (0.6) | 0.312 | 0.138 | 0.0948 | 0.134 |
From this matrix, its actually quite telling that by the numbers, the focal loss seems to be a clear winner. However, when looking at the actual intention of this model, the loss of fidelity on the low frequency notes caused by the focal loss makes it a less desirable option for people trying to transcribe performances. This is because while it is relatively easy to clean up the higher frequencies in post processing, it is much harder to restore data that gets removed through the training process. Therefore, keeping the model using BCE and a threshold of 0.5 helps maintain the most overall structure, and can be cleaned up the best in post processing.
- To isolate the impact of architectural decisions on transcription quality, I tested several CNN configurations against identical training conditions. These 3 experiments had the same loss function, same input data, and same number of epochs. These models were given 10 sample overfit tests to observe how the model learns under controlled conditions. The key variables under investigation were how the convolutional layers were oriented (frequency axis vs. time axis kernels), how much pooling was applied along the frequency dimension, and how the classifier aggregated the final feature map into a per note prediction. Overfitting on a small sample helped remove generalization as a confound, as if the model can’t overfit on 10 samples, the architecture is clearly not working.
- Temporal layer vs. No temporal layer vs. minimal freq pooling
- Time Smearing:
- ![][image8]
- Final stats:
- Epoch 19, Loss: 0.4574480950832367
- logits mean/std: -1.916035771369934 1.1325851678848267
- probs min/mean/max: 0.004231808707118034 0.1756202131509781 0.9200770258903503
- precision: 0.2544688284397125 recall: 0.28620463609695435
- Architecture:
- Conv -> pool -> conv -> pool -> timeConv -> convHead (batch, 88, freqs', W) -> torch mean/max
- We see mostly constant, long term notes, with very little relation to what is happening in the song specifically. The predictions are seemingly agnostic to where in the piece it is, and it will just produce the same lines along the most commonly played notes (across all sample pieces) and play those always. The precision and recall scores, both sitting below 0.30, reflect this: the model isn't making many confident correct predictions, it's making a small number of safe, frequency-agnostic guesses anchored to the training distribution. The temporal layer, rather than adding musical context, effectively gave the model a mechanism to latch onto global note frequency statistics and ignore local spectral evidence
- Frequency smearing:
- ![][image9]
- Final stats:
- Epoch 19, Loss: 0.8720903396606445
- logits mean/std: -2.545652389526367 3.6437182426452637
- probs min/mean/max: 3.438668727540062e-07 0.3067767024040222 0.983948290348053
- precision: 0.0889439582824707 recall: 0.7949493527412415
- Architecture:
- Conv -> pool -> conv -> pool -> conv -> pool -> convHead (B, 88, freqs', W) -> torch mean/max
- Removing the temporal layer entirely and replacing it with a third frequency-axis convolution produced the inverse failure mode. The loss was significantly higher and had a dramatic logit standard deviation. This indicates that the model is far less calibrated, making high confidence predictions that are scattered broadly. The High recall and low precision illustrate this clearly, as the model is activating a large fraction of the 88 output notes at any given frame. The activations do change more along the time axis, not remaining static like in architecture 1, however, the model is not distinguishing the harmonics or neighboring frequencies, so every onset triggers a bunch of false positives.
- Time Smearing:
- QCT: no pooling at all, don’t collapse frequency until very end (264 bins -> 88)
- ![][image4]
- Final Stats:
- Epoch 19, Loss: 0.2656612694263458
- logits mean/std: -2.219456434249878 1.0583772659301758
- probs min/mean/max: 0.004196124617010355 0.1363288164138794 0.9547777771949768
- target mean: 0.027157314121723175
- precision: 0.29081329703330994 recall: 0.559333086013794
- Architecture:
- Conv -> pool -> conv -> conv -> convHead (B, 1, 88, W) -> Squeeze (B, 88, W)
- This approach removed a lot of the intermediate frequency pooling, preserving the full 264 bin CQT frequency resolution for much more of the process and only collapsing at the very end using the squeeze operation. This was because the pooling along the frequency axis lost too much fine-grain frequency information. From the results, we see this to work out quite well. It converges much better than the other models (significantly lower loss), and the probability mean is much closer to the expected mean, proving that it is selectively choosing notes. From the diagram, we can see much better frequency distinction, as it articulates even quick frequency changes. The remaining errors follow a much more predictable pattern, with lower register notes fading since they carry much less spectral power than their harmonics. Additionally, we see the harmonics manifesting themselves in higher notes, which is not super surprising due to how the lower frequency notes will likely have some level of acoustic overlap that increase spectral power at those higher frequencies. These errors are therefore now mostly acoustic-limited, rather than architecturally limited.
- Evaluating this transcription model revealed several systematic failure modes that point to the limits of this CNN architecture and the fundamental complexity of this translation task. As seen in the plots above, the mid-frequency notes were very well modeled, but the high and low frequency notes were not recognized well (over-recognized at high frequencies and under-recognized at low frequencies). This exposed a core generalization problem, where the most commonly played notes across the training set were much more likely to be modeled effectively. This makes sense when thinking about the acoustic essence of the piano, as a single struck note will produce a rich harmonic series, and then the overtones are also combined with the resonance from the piano body and soundboard. Therefore, the CNN sees a spectral bleed from any given note to its harmonics’ frequency bins, making a 1-1 frequency class to mappings difficult to learn without modeling in harmonic structure.
- A second cluster of errors stems from the temporal dimension of piano acoustics. Because the envelope of a piano key has a continuous delay after a sharp attack, the model struggles to account for the amount of sustain for the notes. As the frequency content disappears overtime regardless of how long the note is being held for, the CNN struggles to build those connections. This is probably another reason why lower frequencies were so much harder for the model to predict, as lower region notes tend to be held for longer, and therefore, with less present spectral power in the input, it is harder to connect a low spectral power to a 1 for a held down note later down the line, as at that point in time, there would be very little obvious correlation between that power and the note press. Looking the CQT input/output visualization above, we can see that the bottom notes are very bright when first struck, but there is very little visual cue to see them being held for as long as they were.
- Finally, the dataset introduced systemic bias that was not originally factored into the design. Because all of the data was taken from piano performers, note velocity varied considerably, as that’s one of the greatest sources of expression for a pianist. However, the ground truth outputs were simply 1s and 0s, without factoring in that expression. This resulted in the model being penalized equally for missing a barely audible soft note and for missing a loud one, even though the input evidence was dramatically different in each case. All of these sources contributed some level of error, but there surely could be ways of adapting the model and preprocessing pipeline to fix these errors.
[image4]: <data:image/png;base64,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