Update app.py

app.py

@@ -1,6 +1,23 @@
-from gradio_client import Client
+import subprocess
+subprocess.run(['apt-get', 'update'])
+subprocess.run(['apt-get', 'install', '-y', 'build-essential', 'gawk', 'libasound2-dev', 'libpulse-dev', 'autoconf', 'automake', 'libtool'])
+subprocess.run(['wget', 'https://github.com/espeak-ng/espeak-ng/archive/refs/tags/1.52.0.tar.gz'])
+subprocess.run(['tar', 'xf', '1.52.0.tar.gz'])
+cwd = 'espeak-ng-1.52.0'
+subprocess.run(['./autogen.sh'], cwd=cwd)
+subprocess.run(['./configure'], cwd=cwd)
+subprocess.run(['make'], cwd=cwd)
+subprocess.run(['make', 'install'], cwd=cwd)
+del cwd
+import os
+env = os.environ.copy()
+env['LD_PRELOAD'] = '/usr/local/lib/libespeak-ng.so.1'
+subprocess.run(['espeak-ng', '--version'], env=env)
+from phonemizer.backend.espeak.wrapper import EspeakWrapper
+EspeakWrapper.set_library('/usr/local/lib/libespeak-ng.so.1')
+
+from misaki import en, espeak
 import gradio as gr
-import json
 import os
 import pprint
 import spaces
@@ -9,11 +26,23 @@ import spaces
 def greet(n):
     return f"Hello {zero + n} Tensor"
 
-client = Client('hexgrad/misaki-src', hf_token=os.environ['TOKEN'])
-def predict(text, use_spacy_transformer, british):
-    phonemes, trace, elapsed_cpu_time = client.predict(text=text, trf=use_spacy_transformer, british=british, api_name='/predict')
-    trace = pprint.pformat(json.loads(trace))
-    return phonemes, len(phonemes), trace, elapsed_cpu_time
+fbs = [espeak.EspeakFallback(british=british) for british in (False, True)]
+g2p = [[en.G2P(trf=trf, british=british, fallback=fbs[british]) for british in (False, True)] for trf in (False, True)]
+
+def forward(text, use_spacy_transformer, british):
+    start = time.time()
+    ps, tokens = g2p[use_spacy_transformer][british](text)
+    debug = []
+    for word in tokens:
+        if isinstance(word, list):
+            debug.append([])
+            for t in word:
+                debug[-1].append(t.debug_all())
+        else:
+            debug.append(word.debug_all())
+    trace = pprint.pformat(debug)
+    elapsed_cpu_time = time.time() - start
+    return ps, len(ps), trace, elapsed_cpu_time
 
 with gr.Blocks() as app:
     gr.Markdown('''
@@ -159,7 +188,7 @@ def to_espeak(ps):
 
 demo = gr.TabbedInterface(
     [app, info],
-    ['🔥 Misaki English Preview', 'ℹ️ Phonemes'],
+    ['🔥 Misaki English', 'ℹ️ Phonemes'],
 )
 
 demo.launch()