fix: librosa and requirements

[shigabeev]

Hello!

The existing versions of Numpy and Librosa in the repository are outdated, leading to installation challenges. I've updated the mel extraction function to ensure compatibility with the newer versions of Librosa. Additionally, I've bumped the versions of both Numpy, Cython, phonemizer and Scipy.

The specific change can be found in mel_processing.py

Previously, the code was:

mel = librosa_mel_fn(sampling_rate, n_fft, num_mels, fmin, fmax)

I've modified it to:

mel = librosa_mel_fn(sr=sampling_rate, n_fft=n_fft, n_mels=num_mels, fmin=fmin, fmax=fmax)

Thanks for considering my changes!