| import torch | |
| from transformers import pipeline | |
| from parler_tts import ParlerTTSForConditionalGeneration | |
| from transformers import AutoTokenizer, AutoFeatureExtractor | |
| # ArticMonkey:19.03.24:1700 example of version name in plaintext will be convert into hex using this site -> https://magictool.ai/tool/text-to-hex-converter/ | |
| # Here ArticMonkey is name of version and rest of all is data and time | |
| device = 0 if torch.cuda.is_available() else "cpu" | |
| checkpoint_whisper = "openai/whisper-medium" | |
| pipe = pipeline( | |
| "automatic-speech-recognition", | |
| model=checkpoint_whisper, | |
| device=device, | |
| chunk_length_s=30, | |
| ) | |
| checkpoint_parler = "parler-tts/parler_tts_mini_v0.1" | |
| model_parler = ParlerTTSForConditionalGeneration.from_pretrained(checkpoint_parler).to(device) | |
| tokenizer = AutoTokenizer.from_pretrained(checkpoint_parler) | |
| feature_extractor = AutoFeatureExtractor.from_pretrained(checkpoint_parler) | |
| SAMPLE_RATE = feature_extractor.sampling_rate | |
| SEED = 42 |