| 137 | |
| 138 | |
| 139 | def process_audio_array(audio: Tuple[int, bytes], |
| 140 | env: lmdb.Environment, |
| 141 | channels: int = 1) -> int: |
| 142 | audio_id, audio_samples = audio |
| 143 | buffers = {} |
| 144 | buffers['waveform'] = AudioExample.AudioBuffer( |
| 145 | shape=(channels, int(len(audio_samples) / channels)), |
| 146 | sampling_rate=FLAGS.sampling_rate, |
| 147 | data=audio_samples, |
| 148 | precision=AudioExample.Precision.INT16, |
| 149 | ) |
| 150 | |
| 151 | ae = AudioExample(buffers=buffers) |
| 152 | key = f'{audio_id:08d}' |
| 153 | with env.begin(write=True) as txn: |
| 154 | txn.put( |
| 155 | key.encode(), |
| 156 | ae.SerializeToString(), |
| 157 | ) |
| 158 | return audio_id |
| 159 | |
| 160 | |
| 161 | def process_audio_file(audio: Tuple[int, Tuple[str, float]], |