def create_seed(filename, sample_rate, quantization_channels, window_size=WINDOW): audio, _ = librosa.load(filename, sr=sample_rate, mono=True) audio = audio_reader.trim_silence(audio) quantized = mu_law_encode(audio, quantization_channels) cut_index = tf.cond(tf.size(quantized) < tf.constant(window_size), lambda: tf.size(quantized), lambda: tf.constant(window_size)) return quantized[:cut_index]
def create_seed(filename, sample_rate, quantization_channels, window_size, silence_threshold=SILENCE_THRESHOLD): audio, _ = librosa.load(filename, sr=sample_rate, mono=True) audio = audio_reader.trim_silence(audio, silence_threshold) quantized = mu_law_encode(audio, quantization_channels) cut_index = tf.cond( tf.size(quantized) < tf.constant(window_size), lambda: tf.size(quantized), lambda: tf.constant(window_size)) return quantized
def create_seed(filename, sample_rate, quantization_channels, window_size, scalar_input, silence_threshold=SILENCE_THRESHOLD): audio, _ = librosa.load(filename, sr=sample_rate, mono=True) audio = audio_reader.trim_silence(audio, silence_threshold) if scalar_input: if len(audio) < window_size: return audio else: return audio[:window_size] else: quantized = mu_law_encode(audio, quantization_channels) # 짧으면 짧은 대로 return하는데, padding이라도 해야되지 않나??? cut_index = tf.cond( tf.size(quantized) < tf.constant(window_size), lambda: tf.size(quantized), lambda: tf.constant(window_size)) return quantized[:cut_index]