Jarvis/stt_module.py at main · Couvbat/Jarvis · GitHub

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
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
"""Speech-to-Text module using faster-whisper."""

import numpy as np
from faster_whisper import WhisperModel
from loguru import logger
from config import settings
import tempfile
import soundfile as sf


class STTModule:
    """Speech-to-Text service using faster-whisper."""

    def __init__(self):
        self.model_name = settings.whisper_model
        self.device = settings.whisper_device
        self.compute_type = settings.whisper_compute_type
        self.language = settings.whisper_language
        self.model = None

    def initialize(self):
        """Load the Whisper model."""
        if self.model is not None:
            logger.info("STT model already loaded")
            return

        logger.info(f"Loading Whisper model: {self.model_name} on {self.device}")

        try:
            self.model = WhisperModel(
                self.model_name,
                device=self.device,
                compute_type=self.compute_type
            )
            logger.info("STT model loaded successfully")
        except Exception as e:
            logger.error(f"Failed to load STT model: {e}")
            raise

    def set_language(self, language: str):
        """
        Change the transcription language.

        Args:
            language: Language code ('en', 'fr', 'auto', etc.)
        """
        self.language = language
        logger.info(f"Language changed to: {language}")

    def transcribe(self, audio_data: np.ndarray, sample_rate: int = 16000) -> str:
        """
        Transcribe audio to text.

        Args:
            audio_data: Audio samples as numpy array (int16)
            sample_rate: Sample rate of the audio

        Returns:
            Transcribed text
        """
        if self.model is None:
            raise RuntimeError("STT model not initialized. Call initialize() first.")

        logger.info("Transcribing audio...")

        try:
            # faster-whisper expects float32 audio normalized to [-1, 1]
            if audio_data.dtype == np.int16:
                audio_float = audio_data.astype(np.float32) / 32768.0
            else:
                audio_float = audio_data.astype(np.float32)

            # Transcribe
            segments, info = self.model.transcribe(
                audio_float,
                language=self.language,
                beam_size=5,
                vad_filter=True,
                vad_parameters=dict(min_silence_duration_ms=500)
            )

            # Combine all segments
            transcription = " ".join([segment.text for segment in segments])
            transcription = transcription.strip()

            logger.info(f"Transcription: {transcription}")
            logger.debug(f"Detected language: {info.language} (probability: {info.language_probability:.2f})")

            return transcription

        except Exception as e:
            logger.error(f"Transcription error: {e}")
            raise

    def transcribe_file(self, audio_file: str) -> str:
        """
        Transcribe audio from a file.

        Args:
            audio_file: Path to audio file

        Returns:
            Transcribed text
        """
        if self.model is None:
            raise RuntimeError("STT model not initialized. Call initialize() first.")

        logger.info(f"Transcribing file: {audio_file}")

        try:
            segments, info = self.model.transcribe(
                audio_file,
                language=self.language,
                beam_size=5,
                vad_filter=True
            )

            transcription = " ".join([segment.text for segment in segments])
            transcription = transcription.strip()

            logger.info(f"Transcription: {transcription}")
            return transcription

        except Exception as e:
            logger.error(f"Transcription error: {e}")
            raise