Skip to content

deepgram

deepgram

Deepgram speech-to-text backend (cloud).

Classes

DeepgramSpeechBackend 

DeepgramSpeechBackend(api_key: Optional[str] = None)

Bases: SpeechBackend

Cloud speech-to-text using Deepgram API.

Source code in src/openjarvis/speech/deepgram.py 
24
25
26
27
28
def __init__(self, api_key: Optional[str] = None) -> None:
    self._api_key = api_key or os.environ.get("DEEPGRAM_API_KEY", "")
    self._client = None
    if self._api_key and DeepgramClient is not None:
        self._client = DeepgramClient(self._api_key)
Functions
transcribe 
transcribe(audio: bytes, *, format: str = 'wav', language: Optional[str] = None) -> TranscriptionResult

Transcribe audio using Deepgram's API.

Source code in src/openjarvis/speech/deepgram.py 
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
def transcribe(
    self,
    audio: bytes,
    *,
    format: str = "wav",
    language: Optional[str] = None,
) -> TranscriptionResult:
    """Transcribe audio using Deepgram's API."""
    if self._client is None:
        raise RuntimeError("Deepgram client not initialized (missing API key?)")

    mime_map = {
        "wav": "audio/wav",
        "mp3": "audio/mpeg",
        "ogg": "audio/ogg",
        "flac": "audio/flac",
        "webm": "audio/webm",
        "m4a": "audio/mp4",
    }
    mime_type = mime_map.get(format, "audio/wav")

    options_kwargs: dict = {"model": "nova-2", "smart_format": True}
    if language:
        options_kwargs["language"] = language
    else:
        options_kwargs["detect_language"] = True

    payload = {"buffer": audio, "mimetype": mime_type}

    if PrerecordedOptions is not None:
        options = PrerecordedOptions(**options_kwargs)
    else:
        options = options_kwargs

    response = self._client.listen.rest.v("1").transcribe_file(
        payload, options,
    )

    # Extract transcript from response
    channels = response.results.channels
    if channels and channels[0].alternatives:
        alt = channels[0].alternatives[0]
        text = alt.transcript
        confidence = getattr(alt, "confidence", None)
    else:
        text = ""
        confidence = None

    detected_lang = None
    if channels:
        detected_lang = getattr(channels[0], "detected_language", None)

    duration = getattr(response.metadata, "duration", 0.0)

    return TranscriptionResult(
        text=text,
        language=detected_lang,
        confidence=confidence,
        duration_seconds=duration,
        segments=[],
    )