第１０章 日本語Tactronに基づく音声合成システムの実装のノートブックで、エラーになります。 #43
Description
幸田と申します。
第１０章のノートブックの３２セル目で、以下のようなエラーが出ます。
※実行環境は、Jetson Orinで行っています。
/tmp/ipykernel_11866/3661236322.py:13: FutureWarning: Pass orig_sr=48000, target_sr=16000 as keyword args. From version 0.10 passing these as positional arguments will result in an error
x = librosa.resample(x, _sr, sr)
ValueError Traceback (most recent call last)
File ~/ttslearn/venv/lib/python3.8/site-packages/scipy/signal/windows/_windows.py:2214, in get_window(window, Nx, fftbins)
2213 try:
-> 2214 beta = float(window)
2215 except (TypeError, ValueError) as e:
ValueError: could not convert string to float: 'hanning'
During handling of the above exception, another exception occurred:
KeyError Traceback (most recent call last)
File ~/ttslearn/venv/lib/python3.8/site-packages/scipy/signal/windows/_windows.py:2232, in get_window(window, Nx, fftbins)
2231 try:
-> 2232 winfunc = _win_equiv[winstr]
2233 except KeyError as e:
KeyError: 'hanning'
The above exception was the direct cause of the following exception:
ValueError Traceback (most recent call last)
Cell In [32], line 15
12 x = (x / 32768).astype(np.float64)
13 x = librosa.resample(x, _sr, sr)
---> 15 out_feats = logmelspectrogram(x, sr)
17 # 冒頭と末尾の非音声区間の長さを調整
18 assert "sil" in labels.contexts[0] and "sil" in labels.contexts[-1]
File ~/ttslearn/venv/lib/python3.8/site-packages/ttslearn/dsp.py:310, in logmelspectrogram(y, sr, n_fft, hop_length, win_length, n_mels, fmin, fmax, clip)
307 if n_fft is None:
308 n_fft = next_power_of_2(win_length)
--> 310 S = librosa.stft(
311 y, n_fft=n_fft, hop_length=hop_length, win_length=win_length, window="hanning"
312 )
314 fmin = 0 if fmin is None else fmin
315 fmax = sr // 2 if fmax is None else fmax
File ~/ttslearn/venv/lib/python3.8/site-packages/librosa/util/decorators.py:88, in deprecate_positional_args.._inner_deprecate_positional_args..inner_f(*args, **kwargs)
86 extra_args = len(args) - len(all_args)
87 if extra_args <= 0:
---> 88 return f(*args, **kwargs)
90 # extra_args > 0
91 args_msg = [
92 "{}={}".format(name, arg)
93 for name, arg in zip(kwonly_args[:extra_args], args[-extra_args:])
94 ]
File ~/ttslearn/venv/lib/python3.8/site-packages/librosa/core/spectrum.py:204, in stft(y, n_fft, hop_length, win_length, window, center, dtype, pad_mode)
201 # Check audio is valid
202 util.valid_audio(y, mono=False)
--> 204 fft_window = get_window(window, win_length, fftbins=True)
206 # Pad the window out to n_fft size
207 fft_window = util.pad_center(fft_window, size=n_fft)
File ~/ttslearn/venv/lib/python3.8/site-packages/librosa/util/decorators.py:88, in deprecate_positional_args.._inner_deprecate_positional_args..inner_f(*args, **kwargs)
86 extra_args = len(args) - len(all_args)
87 if extra_args <= 0:
---> 88 return f(*args, **kwargs)
90 # extra_args > 0
91 args_msg = [
92 "{}={}".format(name, arg)
93 for name, arg in zip(kwonly_args[:extra_args], args[-extra_args:])
94 ]
File ~/ttslearn/venv/lib/python3.8/site-packages/librosa/filters.py:1185, in get_window(window, Nx, fftbins)
1180 return window(Nx)
1182 elif isinstance(window, (str, tuple)) or np.isscalar(window):
1183 # TODO: if we add custom window functions in librosa, call them here
-> 1185 return scipy.signal.get_window(window, Nx, fftbins=fftbins)
1187 elif isinstance(window, (np.ndarray, list)):
1188 if len(window) == Nx:
File ~/ttslearn/venv/lib/python3.8/site-packages/scipy/signal/windows/_windows.py:2234, in get_window(window, Nx, fftbins)
2232 winfunc = _win_equiv[winstr]
2233 except KeyError as e:
-> 2234 raise ValueError("Unknown window type.") from e
2236 if winfunc is dpss:
2237 params = (Nx,) + args + (None,)
ValueError: Unknown window type.
なにか、環境やインストールの際のバージョン等に誤りがあるのでしょうか？
恐れ入りますが、よろしくおねがいいたします。