app.py

import gradio as gr
import numpy as np
# Load spectrogram generator
from nemo.collections.tts.models import FastPitchModel
spec_generator = FastPitchModel.from_pretrained(model_name="inOXcrm/German_multispeaker_FastPitch_nemo")


# Load Vocoder
from nemo.collections.tts.models import HifiGanModel
model = HifiGanModel.from_pretrained(model_name="tts_de_hui_hifigan_ft_fastpitch_multispeaker_5")

# Generate audio

def generate_audio(speaker_id, input_txt):
    sr=44100
    parsed = spec_generator.parse(input_txt)
    spectrogram = spec_generator.generate_spectrogram(tokens=parsed, speaker=int(speaker_id))
    audio = model.convert_spectrogram_to_audio(spec=spectrogram)
    audio = audio.to('cpu').detach().numpy()[0]
    audio = audio / np.abs(audio).max()
    return (sr, audio)


gr.Interface(
    generate_audio,
    [
        gr.Textbox(type="text", value=1, label="Speaker ID (1-5)"),
        gr.Textbox(type="text", value="Hallo, wie geht es ihnen?", label="Input Text")
    ],
    "audio",
).launch()