File size: 3,985 Bytes
db7bf9a
 
313814b
 
4bdd7f2
313814b
 
4bdd7f2
db7bf9a
313814b
 
db7bf9a
313814b
 
db7bf9a
313814b
db7bf9a
 
313814b
 
 
db7bf9a
313814b
 
4bdd7f2
 
 
 
db7bf9a
 
 
 
 
 
313814b
4bdd7f2
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
db7bf9a
 
313814b
db7bf9a
313814b
 
4bdd7f2
 
 
e01d72d
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
4bdd7f2
 
 
 
 
 
 
 
 
 
 
 
 
 
db7bf9a
 
 
 
 
 
 
 
 
 
4bdd7f2
db7bf9a
 
 
 
 
 
 
 
 
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
127
128
129
130
131
132
133
134
135
136
137
138
139
from __future__ import annotations

import enum

from faster_whisper.transcribe import Segment, TranscriptionInfo, Word
from pydantic import BaseModel

from speaches import utils
from speaches.core import Transcription


# https://platform.openai.com/docs/api-reference/audio/createTranscription#audio-createtranscription-response_format
class ResponseFormat(enum.StrEnum):
    TEXT = "text"
    JSON = "json"
    VERBOSE_JSON = "verbose_json"
    # VTT = "vtt"
    # SRT = "srt"


# https://platform.openai.com/docs/api-reference/audio/json-object
class TranscriptionJsonResponse(BaseModel):
    text: str

    @classmethod
    def from_segments(cls, segments: list[Segment]) -> TranscriptionJsonResponse:
        return cls(text=utils.segments_text(segments))

    @classmethod
    def from_transcription(
        cls, transcription: Transcription
    ) -> TranscriptionJsonResponse:
        return cls(text=transcription.text)


class WordObject(BaseModel):
    start: float
    end: float
    word: str
    probability: float

    @classmethod
    def from_word(cls, word: Word) -> WordObject:
        return cls(
            start=word.start,
            end=word.end,
            word=word.word,
            probability=word.probability,
        )


class SegmentObject(BaseModel):
    id: int
    seek: int
    start: float
    end: float
    text: str
    tokens: list[int]
    temperature: float
    avg_logprob: float
    compression_ratio: float
    no_speech_prob: float

    @classmethod
    def from_segment(cls, segment: Segment) -> SegmentObject:
        return cls(
            id=segment.id,
            seek=segment.seek,
            start=segment.start,
            end=segment.end,
            text=segment.text,
            tokens=segment.tokens,
            temperature=segment.temperature,
            avg_logprob=segment.avg_logprob,
            compression_ratio=segment.compression_ratio,
            no_speech_prob=segment.no_speech_prob,
        )


# https://platform.openai.com/docs/api-reference/audio/verbose-json-object
class TranscriptionVerboseJsonResponse(BaseModel):
    task: str = "transcribe"
    language: str
    duration: float
    text: str
    words: list[WordObject]
    segments: list[SegmentObject]

    @classmethod
    def from_segment(
        cls, segment: Segment, transcription_info: TranscriptionInfo
    ) -> TranscriptionVerboseJsonResponse:
        return cls(
            language=transcription_info.language,
            duration=segment.end - segment.start,
            text=segment.text,
            words=(
                [WordObject.from_word(word) for word in segment.words]
                if type(segment.words) == list
                else []
            ),
            segments=[SegmentObject.from_segment(segment)],
        )

    @classmethod
    def from_segments(
        cls, segments: list[Segment], transcription_info: TranscriptionInfo
    ) -> TranscriptionVerboseJsonResponse:
        return cls(
            language=transcription_info.language,
            duration=transcription_info.duration,
            text=utils.segments_text(segments),
            segments=[SegmentObject.from_segment(segment) for segment in segments],
            words=[
                WordObject.from_word(word)
                for word in utils.words_from_segments(segments)
            ],
        )

    @classmethod
    def from_transcription(
        cls, transcription: Transcription
    ) -> TranscriptionVerboseJsonResponse:
        return cls(
            language="english",  # FIX: hardcoded
            duration=transcription.duration,
            text=transcription.text,
            words=[
                WordObject(
                    start=word.start,
                    end=word.end,
                    word=word.text,
                    probability=word.probability,
                )
                for word in transcription.words
            ],
            segments=[],  # FIX: hardcoded
        )