From 03b5b8fef082884e726fa08777d6b91463bee638 Mon Sep 17 00:00:00 2001 From: Pavel Date: Tue, 7 Nov 2023 04:20:34 +0700 Subject: [PATCH] Update description of n_mels according to changes mel_filters function --- whisper/audio.py | 4 ++-- 1 file changed, 2 insertions(+), 2 deletions(-) diff --git a/whisper/audio.py b/whisper/audio.py index cf6c66a..06c30d2 100644 --- a/whisper/audio.py +++ b/whisper/audio.py @@ -122,7 +122,7 @@ def log_mel_spectrogram( The path to audio or either a NumPy array or Tensor containing the audio waveform in 16 kHz n_mels: int - The number of Mel-frequency filters, only 80 is supported + The number of Mel-frequency filters, only 80 or 128 is supported padding: int Number of zero samples to pad to the right @@ -132,7 +132,7 @@ def log_mel_spectrogram( Returns ------- - torch.Tensor, shape = (80, n_frames) + torch.Tensor, shape = (80, n_frames) or (128, n_frames), depends on n_mels value A Tensor that contains the Mel spectrogram """ if not torch.is_tensor(audio):