Talk:Audio inpainting

Citation
From a lot of citation is missing a proper journal field Ilgeco1995 (talk) 13:31, 7 July 2023 (UTC)


 * I'm agree with @Ilgeco1995, moreover in the book-reference n.7 the number page is missing. Moreover, also categories are missing. Giacomo Speroni (talk) 13:42, 7 July 2023 (UTC)
 * Thank you for noting this. I am editing the references, adding the missing fields. Fdrc.Mtll (talk) 14:37, 7 July 2023 (UTC)

Equation in the "Definition" section
I have checked the references and I do not find the equation that you post. I understand the mathematical meaning, but I think that if is a reformulation of a more complicated equation it should be indicated by a note.

Moreover, if it is correct I think that you can be more precise about this formula. Maybe you can add a more clear definition of $$ \tilde{x} $$ and also on $$ \hat{x}^* $$. Giacomo Speroni (talk) 13:54, 7 July 2023 (UTC)


 * Thank you for the comment. The formula can be found in references 11 and 12. Nonetheless, I agree that it needs a clearer explanation. I will edit the section. Fdrc.Mtll (talk) 14:40, 7 July 2023 (UTC)

Review
Nice work in summing up the different approaches to the problem and the different solutions that are widely adopted. I'll check the references for sure as I have a practical problem of audio denoising at hand that could be suitably solved with the proposed techniques. Primoh (talk) 14:12, 7 July 2023 (UTC)

Review
Nice work, really well done. A page that Wikipedia was missing! Daniele leonzio (talk) 15:19, 7 July 2023 (UTC)

Observations and suggestions for improvements
The following observations and suggestions for improvements were collected, following an expert review of the article within the Science, Technology, Society and Wikipedia course at the Politecnico di Milano, in July 2023. The current formulation of the minimization problem suggests that $\tilde{x}$ is compared to $\mathbf{m} \circ \mathbf{\hat{x}}$, i.e., the masked version of \mathbf{\hat{x}} (see the definition $L( \mathbf{m} \circ \mathbf{\hat{x}}, \mathbf{\tilde{x}} )$) The text suggests that only the reliable part of the signal is used in the minimization problem, which makes sense. I would then consider making this explicit in the mathematical formulation changing the current $L$ definition with $L( \mathbf{m} \circ \mathbf{\hat{x}}, \mathbf{m} \circ \mathbf{\tilde{x}} )$, where both $\mathbf{\hat{x}$ and $\mathbf{\tilde{x}}$ are masked in order to avoid comparing zeros with signal samples.

--Aandurro (talk) 12:00, 28 July 2023 (UTC)


 * Thank you for your comment. It is true that only the reliable part of the signal is used in the minimization problem. However, writing it as $L(\mathbf{m} \circ \mathbf{\hat{x}}, \mathbf{m} \circ \mathbf{\tilde{x}})$ would be an error, since $\mathbf{\tilde{x}}$ is already defined as $\mathbf{m} \circ \mathbf{x}$ and thus only contains the reliable part of the signal. Fdrc.Mtll (talk) 18:08, 29 July 2023 (UTC)