File size: 2,391 Bytes
a9bf4b2
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
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
import base64
import os
import time
from dataclasses import dataclass
from datetime import timedelta

import banana_dev as banana
import gradio as gr
from loguru import logger
from pydub import AudioSegment

api_key = os.environ["BANANA_API_KEY"]
model_key = os.environ["BANANA_MODEL_KEY"]
password = os.environ["PASSWORD"]


def transcribe(audio=None, url=None):
    if audio:
        audio_b64 = base64.b64encode(audio.export().read()).decode("ascii")
        payload = {"audio_b64": audio_b64}
    else:
        payload = {"url": url}
    response = banana.run(api_key, model_key, payload)
    print(response)

    if "error" in response:
        raise gr.Error(response["error"])

    # TODO: not sure why response dict contains multiple model outputs
    return response["modelOutputs"][0]


def run_demo(password, url, file_upload):
    if password not in [os.environ["PASSWORD"], os.environ["ROOT_PASSWORD"]]:
        raise gr.Error("Der Zugriffscode ist falsch.")

    if (url is not None) and (file_upload is not None):
        logger.warning(
            "Achtung: Sie haben sowohl eine URL angegeben als auch eine Datei hochgeladen."
            " Wir verwenden nur die Datei, die Sie hochgeladen haben."
        )

    elif (url is None) and (file_upload is None):
        raise gr.Error(
            "Sie müssen entweder eine URL angeben oder eine Datei hochladen."
        )

    start = time.time()

    if file_upload is not None:
        cutoff = None if password == os.environ["ROOT_PASSWORD"] else 60_000
        audio = AudioSegment.from_file(file_upload, format="mp3")[:cutoff]
        transcription = transcribe(audio=audio, url=None)
    else:
        transcription = transcribe(audio=None, url=url)

    logger.info(f"transcription took {time.time()-start:.3f}s")
    return transcription["text"]


demo = gr.Interface(
    fn=run_demo,
    inputs=[
        # gr.Textbox(label="Email", type="email"),
        gr.Textbox(label="Zugriffscode (siehe oben)"),
        # gr.Audio(source="microphone", type="filepath", label="Aufnehmen"),
        gr.Textbox(label="URL (z.B. YouTube-Video, Dropbox-Datei, etc.)"),
        gr.Audio(source="upload", type="filepath", label="Datei hochladen"),
    ],
    outputs=gr.Textbox(label="Automatisches Transkript"),
    allow_flagging="never",
    css="footer {visibility: hidden} .meta-text {visibility: hidden}",
)

demo.launch()