Create app.py

app.py

@@ -0,0 +1,95 @@
+import gradio as gr
+import os
+os.system('cd monotonic_align && python setup.py build_ext --inplace && cd ..')
+
+import json
+import math
+import torch
+from torch import nn
+from torch.nn import functional as F
+from torch.utils.data import DataLoader
+
+import commons
+import utils
+from data_utils import TextAudioLoader, TextAudioCollate, TextAudioSpeakerLoader, TextAudioSpeakerCollate
+from models import SynthesizerTrn
+#from text.symbols import symbols
+from wa_graphemes.symbols import symbols
+from text import text_to_sequence
+from scipy.io.wavfile import write
+
+def get_text(text, hps):
+    text_norm = text_to_sequence(text, hps.data.text_cleaners)
+    if hps.data.add_blank:
+        text_norm = commons.intersperse(text_norm, 0)
+    text_norm = torch.LongTensor(text_norm)
+    return text_norm
+
+def load_model(model_path, hps):
+    net_g = SynthesizerTrn(
+        len(symbols),
+        hps.data.filter_length // 2 + 1,
+        hps.train.segment_size // hps.data.hop_length,
+        n_speakers=hps.data.n_speakers,
+        **hps.model)
+    _ = net_g.eval()
+    _ = utils.load_checkpoint(model_path, net_g, None)
+    return net_g
+
+#hps = utils.get_hparams_from_file("configs/vctk_base.json")
+hps = utils.get_hparams_from_file("wa_graphemes/config.json")
+
+# Define a dictionary to store the model paths for each tab
+model_paths = {
+    "Graphemes": "wa_graphemes/G_258000.pth"
+}
+
+# Load the initial model
+net_g = load_model(model_paths["Graphemes"], hps)
+
+def tts(text, speaker_id, tab_name):
+    global net_g
+    net_g = load_model(model_paths[tab_name], hps)
+    sid = torch.LongTensor([speaker_id])  # speaker identity
+    stn_tst = get_text(text, hps)
+
+    with torch.no_grad():
+        x_tst = stn_tst.unsqueeze(0)
+        x_tst_lengths = torch.LongTensor([stn_tst.size(0)])
+        audio = net_g.infer(x_tst, x_tst_lengths, sid=sid, noise_scale=.667, noise_scale_w=0.8, length_scale=1)[0][
+            0, 0].data.float().numpy()
+    return "Success", (hps.data.sampling_rate, audio)
+
+def create_tab(tab_name):
+    with gr.TabItem(tab_name):
+        gr.Markdown(f"### {tab_name} TTS Model")
+        tts_input1 = gr.TextArea(label="Text in Walloon on IPA phonemes", value="")
+        tts_input2 = gr.Dropdown(label="Speaker", choices=["Male", "Female"], type="index", value="Male")
+        tts_submit = gr.Button("Generate", variant="primary")
+        tts_output1 = gr.Textbox(label="Message")
+        tts_output2 = gr.Audio(label="Output")
+        tts_submit.click(lambda text, speaker_id: tts(text, speaker_id, tab_name), [tts_input1, tts_input2], [tts_output1, tts_output2])
+
+app = gr.Blocks()
+with app:
+    gr.Markdown(
+        """
+        # First Text to Speech (TTS) for Walloon
+        Based on VITS (https://github.com/jaywalnut310/vits).
+        Write the text in graphemes. For faster inference speed it is recommended to use short sentences.
+        """
+    )
+    with gr.Tabs():
+        create_tab("Phonemes_finetuned")
+
+    gr.Markdown(
+        """
+        ### Examples
+        | Input Text | Speaker | Input Method  |
+        |------------|---------|---------------|
+        | li biːç ɛ l sɔlja ɛstẽ ki s maʁɡajẽ pɔ sawɛ kiː ski , dɛ døː , ɛstøː l py fwaʁ . m ɛ̃ s koː la , la k i vɛjɛ õ tsminɔː k aʁivef pjim pjam , d ɛ̃ õ bja nuː tsoː paltɔ .  | Female | Phonemes |
+        | Li bijhe et l’ solea estént ki s’ margayént po sawè kî çki, des deus, esteut l’ pus foirt. Mins ç’ côp la, la k’ i veyèt on tchminåd k' arivéve pyim piam, dins on bea noû tchôd paltot.  | Male | Graphemes |
+        """
+    )
+
+app.launch()