File size: 1,115 Bytes
a197707 3a02e1e a9270aa a197707 3a02e1e a032b7b 477ce7b a032b7b 1a34264 3a02e1e a8fe0f0 3a02e1e a9270aa |
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 |
---
license: apache-2.0
language:
- fi
tags:
- speech-recognition
- whisper
---
Example how to use with WhisperX (https://github.com/m-bain/whisperX)
```python
import whisperx
device = "cuda"
audio_file = "oma_nauhoitus_16kHz.wav"
batch_size = 16 # reduce if low on GPU mem
compute_type = "float16" # change to "int8" if low on GPU mem (may reduce accuracy)
# 1. Transcribe with original whisper (batched)
model = whisperx.load_model("Finnish-NLP/whisper-large-finnish-v3-ct2", device, compute_type=compute_type)
audio = whisperx.load_audio(audio_file)
result = model.transcribe(audio, batch_size=batch_size)
print(result["segments"]) # before alignment
```
How to use in Python with faster-whisper (https://github.com/SYSTRAN/faster-whisper)
```python
import faster_whisper
model = faster_whisper.WhisperModel("Finnish-NLP/whisper-large-finnish-v3-ct2")
print("model loaded")
segments, info = model.transcribe(audio_path, word_timestamps=True, beam_size=5, language="fi")
for segment in segments:
for word in segment.words:
print("[%.2fs -> %.2fs] %s" % (word.start, word.end, word.word))
``` |