""")
with gr.Tabs():
# ── TAB 1: Speech to Text ──────────────────────────────────────────
with gr.TabItem("🎤 Speech → Text", elem_classes=["tabs-container"]):
gr.Markdown("""
**Upload or record audio** and Eburon Audio will transcribe it with high accuracy.
Supports multiple languages, handles noisy inputs, and can detect the language automatically.
""")
with gr.Row():
with gr.Column(scale=1):
stt_audio = gr.Audio(
label="Audio Input",
sources=["microphone", "upload"],
type="filepath",
elem_classes=["audio-component"],
)
stt_language = gr.Dropdown(
choices=LANGUAGES,
value="Auto-detect",
label="Language",
)
stt_btn = gr.Button("✨ Transcribe", variant="primary")
with gr.Column(scale=1):
stt_output = gr.Textbox(
label="Transcription",
lines=12,
placeholder="Your transcribed text will appear here...",
)
stt_btn.click(
fn=transcribe_handler,
inputs=[stt_audio, stt_language],
outputs=stt_output,
)
# ── TAB 2: Text to Speech ──────────────────────────────────────────
with gr.TabItem("🔊 Text → Speech", elem_classes=["tabs-container"]):
gr.Markdown("""
**Type text** and Eburon Audio converts it into natural speech.
Optionally paste a **Voice ID** from the Voice Cloning tab to use your own cloned voice.
""")
with gr.Row():
with gr.Column(scale=1):
tts_text = gr.Textbox(
label="Text to speak",
lines=8,
placeholder="Enter text here (max ~300 words for best results). Avoid markdown or special characters.",
value="And that is what makes this moment so important.\n\nBecause we are no longer in the phase where AI is only a spectacle.\nWe are entering the phase where it must become dependable.\nIntegrated.\nEfficient.\nAnd truly beneficial.\n\nThe projects that matter now will be the ones that combine vision with grounded execution.\nThe ones that understand cost as well as capability.\nLatency as well as intelligence.\nHuman need as well as model performance.",
)
with gr.Row():
tts_voice_id = gr.Dropdown(
label="Select a Voice or Your Session Clones",
choices=INITIAL_VOICES,
value=INITIAL_VOICES[0][1] if INITIAL_VOICES else None,
allow_custom_value=True,
scale=3,
)
voices_btn = gr.Button("🔄 Refresh List", size="sm", scale=1)
voices_list_out = gr.Markdown(visible=False)
tts_ref_audio = gr.Audio(
label="OR: Reference Audio (Set voice tone instantly)",
sources=["upload", "microphone"],
type="filepath",
)
tts_format = gr.Dropdown(
choices=["mp3", "wav", "flac", "opus"],
value="mp3",
label="Audio Format",
)
tts_btn = gr.Button("🎵 Generate Speech", variant="primary")
with gr.Column(scale=1):
tts_audio_out = gr.Audio(
label="Generated Audio",
type="filepath",
elem_classes=["audio-component"],
)
tts_status = gr.Markdown(elem_classes=["status-text"])
tts_btn.click(
fn=synthesize_handler,
inputs=[tts_text, tts_voice_id, tts_ref_audio, tts_format],
outputs=[tts_audio_out, tts_status],
)
voices_btn.click(
fn=lambda: gr.update(choices=get_voice_choices()),
inputs=[],
outputs=tts_voice_id,
)
# ── TAB 3: Voice Cloning ───────────────────────────────────────────
with gr.TabItem("🧬 Voice Cloning", elem_classes=["tabs-container"]):
gr.Markdown("""
**Clone any voice** by uploading a short audio sample (10–60 seconds recommended).
The system will save it as a reusable voice. Copy the Voice ID and paste it in the Text-to-Speech tab.
> ⚠️ Only clone voices with explicit consent.
""")
with gr.Row():
with gr.Column(scale=1):
clone_audio = gr.Audio(
label="Voice Sample (upload or record)",
sources=["microphone", "upload"],
type="filepath",
elem_classes=["audio-component"],
)
clone_url = gr.Textbox(
label="OR: Media URL (TikTok, Twitter, or direct .MP3/.WAV link)",
placeholder="https://...link_to_audio_or_video...",
)
clone_name = gr.Textbox(
label="Voice Name",
placeholder="e.g. eburon-assistant-voice",
)
clone_gender = gr.Dropdown(
choices=["Female", "Male"],
value="Female",
label="Gender",
)
clone_langs = gr.Textbox(
label="Languages (comma-separated)",
value="en",
placeholder="en, fr, es",
)
clone_btn = gr.Button("🧬 Clone Voice", variant="primary")
with gr.Column(scale=1):
clone_result = gr.Markdown(
value="Your new Voice ID will appear here after cloning.",
elem_classes=["status-text"],
)
clone_btn.click(
fn=clone_handler,
inputs=[clone_audio, clone_url, clone_name, clone_gender, clone_langs],
outputs=[clone_result, tts_voice_id],
)
# ── TAB 4: Sample ──────────────────────────────────────────────────
with gr.TabItem("🎧 Sample", elem_classes=["tabs-container"]):
gr.Markdown("""
**Preview the sample audio** below.
This is rendered as a native playable audio sample without affecting the existing app flow.
""")
gr.HTML(f"""