Outdated kaiser Import in ESPnet2 TTS Colab Notebook (pqmf.py Still Using Old SciPy Import)

Describe the Bug: In the ESPnet2 TTS Google Colab notebook (link), an outdated import statement in pqmf.py causes a failure during execution.

The Colab notebook still uses an older version of kaiser, whereas the latest ESPnet2 repository includes an updated import:

try:
    from scipy.signal import kaiser
except ImportError:
    from scipy.signal.windows import kaiser

This update was necessary because SciPy 1.1.0 moved kaiser from scipy.signal to scipy.signal.windows (source).

The main ESPnet2 codebase already reflects this change, but the Colab notebook does not.

Basic Environments:

• OS information: Google Colab (latest)
• python version: 3.11 (Google Colab default)
• espnet version: Latest version from Colab install
• pytorch version: 2.1.0 (Google Colab default)

To Reproduce: Steps to reproduce the behavior:

1. Open the ESPnet2 TTS Colab Demo.
2. Run the notebook cells as instructed.
3. Error occurs due to incorrect SciPy import.

Error Logs:

ImportError: cannot import name 'kaiser' from 'scipy.signal' ([/usr/local/lib/python3](https://colab.research.google.com/github/espnet/notebook/blob/master/ESPnet2/Demo/TTS/tts_realtime_demo.ipynb#).11[/dist-packages/scipy/signal](https://colab.research.google.com/github/espnet/notebook/blob/master/ESPnet2/Demo/TTS/tts_realtime_demo.ipynb#)/__init__.py)

Proposed Fix:

• Update the Google Colab notebook to reflect the fix already present in pqmf.py.
• Should I submit a PR for this?