Fix Gradio 6.0 theme/css warning and refresh speaker status after lock.
Co-authored-by: Cursor <cursoragent@cursor.com>
This commit is contained in:
dekun
@@ -1,378 +1,386 @@
|
|||||||
"""
|
"""
|
||||||
Trading Studio — 自动化交易复盘视频配音系统
|
Trading Studio — 自动化交易复盘视频配音系统
|
||||||
Gradio Web 中控：音色锁定 → Whisper 识别 → Gemma4 润色 → ChatTTS 合成
|
Gradio Web 中控：音色锁定 → Whisper 识别 → Gemma4 润色 → ChatTTS 合成
|
||||||
"""
|
"""
|
||||||
|
|
||||||
from __future__ import annotations
|
from __future__ import annotations
|
||||||
|
|
||||||
import logging
|
import logging
|
||||||
import shutil
|
import shutil
|
||||||
import sys
|
import sys
|
||||||
import uuid
|
import uuid
|
||||||
from pathlib import Path
|
from pathlib import Path
|
||||||
|
|
||||||
import gradio as gr
|
import gradio as gr
|
||||||
|
|
||||||
from config import (
|
from config import (
|
||||||
GIT_REPO_URL,
|
GIT_REPO_URL,
|
||||||
HOST,
|
HOST,
|
||||||
MODEL_NAME,
|
MODEL_NAME,
|
||||||
OLLAMA_URL,
|
OLLAMA_URL,
|
||||||
PORT,
|
PORT,
|
||||||
SPEAKER_EMB_PATH,
|
SPEAKER_EMB_PATH,
|
||||||
UPLOAD_DIR,
|
UPLOAD_DIR,
|
||||||
)
|
)
|
||||||
from llm_service import check_ollama_health, polish_text
|
from llm_service import check_ollama_health, polish_text
|
||||||
from tts_service import generate_voice, save_fixed_speaker, speaker_is_ready
|
from tts_service import generate_voice, save_fixed_speaker, speaker_is_ready
|
||||||
from whisper_service import transcribe_audio
|
from whisper_service import transcribe_audio
|
||||||
|
|
||||||
# ---------------------------------------------------------------------------
|
# ---------------------------------------------------------------------------
|
||||||
# 日志
|
# 日志
|
||||||
# ---------------------------------------------------------------------------
|
# ---------------------------------------------------------------------------
|
||||||
logging.basicConfig(
|
logging.basicConfig(
|
||||||
level=logging.INFO,
|
level=logging.INFO,
|
||||||
format="%(asctime)s [%(levelname)s] %(name)s: %(message)s",
|
format="%(asctime)s [%(levelname)s] %(name)s: %(message)s",
|
||||||
handlers=[
|
handlers=[
|
||||||
logging.StreamHandler(sys.stdout),
|
logging.StreamHandler(sys.stdout),
|
||||||
logging.FileHandler("trading_studio.log", encoding="utf-8"),
|
logging.FileHandler("trading_studio.log", encoding="utf-8"),
|
||||||
],
|
],
|
||||||
)
|
)
|
||||||
logger = logging.getLogger("trading_studio")
|
logger = logging.getLogger("trading_studio")
|
||||||
|
|
||||||
# ---------------------------------------------------------------------------
|
# ---------------------------------------------------------------------------
|
||||||
# 全局 UI 状态（Gradio State）
|
# 全局 UI 状态（Gradio State）
|
||||||
# ---------------------------------------------------------------------------
|
# ---------------------------------------------------------------------------
|
||||||
# raw_transcript / polished_script 在流水线中传递
|
# raw_transcript / polished_script 在流水线中传递
|
||||||
|
|
||||||
|
|
||||||
def _save_upload(upload_file) -> str | None:
|
def _save_upload(upload_file) -> str | None:
|
||||||
"""将 Gradio 上传文件复制到本地 uploads 目录，返回持久化路径。"""
|
"""将 Gradio 上传文件复制到本地 uploads 目录，返回持久化路径。"""
|
||||||
if upload_file is None:
|
if upload_file is None:
|
||||||
return None
|
return None
|
||||||
|
|
||||||
src = Path(upload_file)
|
src = Path(upload_file)
|
||||||
if not src.exists():
|
if not src.exists():
|
||||||
return None
|
return None
|
||||||
|
|
||||||
dest = UPLOAD_DIR / f"{uuid.uuid4().hex}_{src.name}"
|
dest = UPLOAD_DIR / f"{uuid.uuid4().hex}_{src.name}"
|
||||||
shutil.copy2(src, dest)
|
shutil.copy2(src, dest)
|
||||||
return str(dest)
|
return str(dest)
|
||||||
|
|
||||||
|
|
||||||
# ---------------------------------------------------------------------------
|
# ---------------------------------------------------------------------------
|
||||||
# 模块 1：音色锁定
|
# 模块 1：音色锁定
|
||||||
# ---------------------------------------------------------------------------
|
# ---------------------------------------------------------------------------
|
||||||
def ui_lock_speaker(audio_file, sample_transcript: str) -> str:
|
def ui_lock_speaker(audio_file, sample_transcript: str) -> tuple[str, str]:
|
||||||
"""【音色锁定】从参考人声提取并保存 Speaker Embedding。"""
|
"""【音色锁定】从参考人声提取并保存 Speaker Embedding。"""
|
||||||
path = _save_upload(audio_file)
|
path = _save_upload(audio_file)
|
||||||
if not path:
|
if not path:
|
||||||
return "请上传 10-30 秒干净参考人声（wav/mp3 均可）。"
|
return "请上传 10-30 秒干净参考人声（wav/mp3 均可）。", ui_speaker_status()
|
||||||
|
|
||||||
ok, msg = save_fixed_speaker(path, sample_transcript or "")
|
ok, msg = save_fixed_speaker(path, sample_transcript or "")
|
||||||
return msg if ok else f"❌ {msg}"
|
result = msg if ok else f"❌ {msg}"
|
||||||
|
return result, ui_speaker_status()
|
||||||
|
|
||||||
def ui_speaker_status() -> str:
|
|
||||||
"""刷新音色状态。"""
|
def ui_speaker_status() -> str:
|
||||||
ok, msg = speaker_is_ready()
|
"""刷新音色状态。"""
|
||||||
return f"✅ {msg}" if ok else f"⚠️ {msg}"
|
ok, msg = speaker_is_ready()
|
||||||
|
return f"✅ {msg}" if ok else f"⚠️ {msg}"
|
||||||
|
|
||||||
# ---------------------------------------------------------------------------
|
|
||||||
# 模块 2：音频极速识别
|
# ---------------------------------------------------------------------------
|
||||||
# ---------------------------------------------------------------------------
|
# 模块 2：音频极速识别
|
||||||
def ui_transcribe(audio_file) -> tuple[str, str]:
|
# ---------------------------------------------------------------------------
|
||||||
"""【Whisper 识别】返回 (转写文本, 状态日志)。"""
|
def ui_transcribe(audio_file) -> tuple[str, str]:
|
||||||
path = _save_upload(audio_file)
|
"""【Whisper 识别】返回 (转写文本, 状态日志)。"""
|
||||||
if not path:
|
path = _save_upload(audio_file)
|
||||||
return "", "请上传待识别的碎碎念录音。"
|
if not path:
|
||||||
|
return "", "请上传待识别的碎碎念录音。"
|
||||||
ok, result = transcribe_audio(path)
|
|
||||||
if ok:
|
ok, result = transcribe_audio(path)
|
||||||
return result, f"✅ 识别完成，共 {len(result)} 字。"
|
if ok:
|
||||||
return "", f"❌ {result}"
|
return result, f"✅ 识别完成，共 {len(result)} 字。"
|
||||||
|
return "", f"❌ {result}"
|
||||||
|
|
||||||
# ---------------------------------------------------------------------------
|
|
||||||
# 模块 3：Gemma4 纪律审判
|
# ---------------------------------------------------------------------------
|
||||||
# ---------------------------------------------------------------------------
|
# 模块 3：Gemma4 纪律审判
|
||||||
def ui_polish(raw_text: str) -> tuple[str, str]:
|
# ---------------------------------------------------------------------------
|
||||||
"""【LLM 润色】对转写稿进行严厉自我反思式润色。"""
|
def ui_polish(raw_text: str) -> tuple[str, str]:
|
||||||
if not raw_text or not raw_text.strip():
|
"""【LLM 润色】对转写稿进行严厉自我反思式润色。"""
|
||||||
return "", "请先完成语音识别或手动粘贴转写文本。"
|
if not raw_text or not raw_text.strip():
|
||||||
|
return "", "请先完成语音识别或手动粘贴转写文本。"
|
||||||
ok, result = polish_text(raw_text)
|
|
||||||
if ok:
|
ok, result = polish_text(raw_text)
|
||||||
return result, f"✅ Gemma4 润色完成，共 {len(result)} 字。"
|
if ok:
|
||||||
return "", f"❌ {result}"
|
return result, f"✅ Gemma4 润色完成，共 {len(result)} 字。"
|
||||||
|
return "", f"❌ {result}"
|
||||||
|
|
||||||
def ui_check_ollama() -> str:
|
|
||||||
"""检测远程 Ollama 节点状态。"""
|
def ui_check_ollama() -> str:
|
||||||
ok, msg = check_ollama_health()
|
"""检测远程 Ollama 节点状态。"""
|
||||||
return f"✅ {msg}" if ok else f"❌ {msg}"
|
ok, msg = check_ollama_health()
|
||||||
|
return f"✅ {msg}" if ok else f"❌ {msg}"
|
||||||
|
|
||||||
# ---------------------------------------------------------------------------
|
|
||||||
# 模块 4：ChatTTS 音频合成
|
# ---------------------------------------------------------------------------
|
||||||
# ---------------------------------------------------------------------------
|
# 模块 4：ChatTTS 音频合成
|
||||||
def ui_synthesize(polished_text: str) -> tuple[str | None, str]:
|
# ---------------------------------------------------------------------------
|
||||||
"""【TTS 合成】生成最终 wav 配音文件。"""
|
def ui_synthesize(polished_text: str) -> tuple[str | None, str]:
|
||||||
if not polished_text or not polished_text.strip():
|
"""【TTS 合成】生成最终 wav 配音文件。"""
|
||||||
return None, "请先完成 Gemma4 润色。"
|
if not polished_text or not polished_text.strip():
|
||||||
|
return None, "请先完成 Gemma4 润色。"
|
||||||
ok, msg, wav_path = generate_voice(polished_text)
|
|
||||||
if ok and wav_path:
|
ok, msg, wav_path = generate_voice(polished_text)
|
||||||
return wav_path, f"✅ {msg}"
|
if ok and wav_path:
|
||||||
return None, f"❌ {msg}"
|
return wav_path, f"✅ {msg}"
|
||||||
|
return None, f"❌ {msg}"
|
||||||
|
|
||||||
# ---------------------------------------------------------------------------
|
|
||||||
# 一键流水线
|
# ---------------------------------------------------------------------------
|
||||||
# ---------------------------------------------------------------------------
|
# 一键流水线
|
||||||
def ui_full_pipeline(
|
# ---------------------------------------------------------------------------
|
||||||
audio_file,
|
def ui_full_pipeline(
|
||||||
skip_polish: bool,
|
audio_file,
|
||||||
manual_raw: str,
|
skip_polish: bool,
|
||||||
) -> tuple[str, str, str | None, str]:
|
manual_raw: str,
|
||||||
"""
|
) -> tuple[str, str, str | None, str]:
|
||||||
串联执行：识别 → 润色（可跳过）→ 合成。
|
"""
|
||||||
返回 (raw, polished, wav_path, log)
|
串联执行：识别 → 润色（可跳过）→ 合成。
|
||||||
"""
|
返回 (raw, polished, wav_path, log)
|
||||||
logs: list[str] = []
|
"""
|
||||||
|
logs: list[str] = []
|
||||||
# Step 1: 识别
|
|
||||||
if manual_raw and manual_raw.strip():
|
# Step 1: 识别
|
||||||
raw = manual_raw.strip()
|
if manual_raw and manual_raw.strip():
|
||||||
logs.append(f"使用手动输入转写稿（{len(raw)} 字）。")
|
raw = manual_raw.strip()
|
||||||
else:
|
logs.append(f"使用手动输入转写稿（{len(raw)} 字）。")
|
||||||
path = _save_upload(audio_file)
|
else:
|
||||||
if not path:
|
path = _save_upload(audio_file)
|
||||||
return "", "", None, "❌ 请上传录音或手动填写转写文本。"
|
if not path:
|
||||||
ok, result = transcribe_audio(path)
|
return "", "", None, "❌ 请上传录音或手动填写转写文本。"
|
||||||
if not ok:
|
ok, result = transcribe_audio(path)
|
||||||
return "", "", None, f"❌ 识别失败: {result}"
|
if not ok:
|
||||||
raw = result
|
return "", "", None, f"❌ 识别失败: {result}"
|
||||||
logs.append(f"✅ Whisper 识别完成（{len(raw)} 字）。")
|
raw = result
|
||||||
|
logs.append(f"✅ Whisper 识别完成（{len(raw)} 字）。")
|
||||||
# Step 2: 润色
|
|
||||||
if skip_polish:
|
# Step 2: 润色
|
||||||
polished = raw
|
if skip_polish:
|
||||||
logs.append("已跳过 LLM 润色，直接使用原文。")
|
polished = raw
|
||||||
else:
|
logs.append("已跳过 LLM 润色，直接使用原文。")
|
||||||
ok, result = polish_text(raw)
|
else:
|
||||||
if not ok:
|
ok, result = polish_text(raw)
|
||||||
return raw, "", None, f"❌ 润色失败: {result}\n" + "\n".join(logs)
|
if not ok:
|
||||||
polished = result
|
return raw, "", None, f"❌ 润色失败: {result}\n" + "\n".join(logs)
|
||||||
logs.append(f"✅ Gemma4 润色完成（{len(polished)} 字）。")
|
polished = result
|
||||||
|
logs.append(f"✅ Gemma4 润色完成（{len(polished)} 字）。")
|
||||||
# Step 3: 合成
|
|
||||||
ok, msg, wav_path = generate_voice(polished)
|
# Step 3: 合成
|
||||||
if not ok:
|
ok, msg, wav_path = generate_voice(polished)
|
||||||
return raw, polished, None, f"❌ 合成失败: {msg}\n" + "\n".join(logs)
|
if not ok:
|
||||||
|
return raw, polished, None, f"❌ 合成失败: {msg}\n" + "\n".join(logs)
|
||||||
logs.append(f"✅ {msg}")
|
|
||||||
return raw, polished, wav_path, "\n".join(logs)
|
logs.append(f"✅ {msg}")
|
||||||
|
return raw, polished, wav_path, "\n".join(logs)
|
||||||
|
|
||||||
# ---------------------------------------------------------------------------
|
|
||||||
# Gradio 界面
|
# ---------------------------------------------------------------------------
|
||||||
# ---------------------------------------------------------------------------
|
# Gradio 界面
|
||||||
CUSTOM_CSS = """
|
# ---------------------------------------------------------------------------
|
||||||
/* 硬核暗黑科技风 */
|
CUSTOM_CSS = """
|
||||||
.gradio-container {
|
/* 硬核暗黑科技风 */
|
||||||
background: linear-gradient(160deg, #0a0a0f 0%, #12121a 40%, #0d0d12 100%) !important;
|
.gradio-container {
|
||||||
color: #c8c8d0 !important;
|
background: linear-gradient(160deg, #0a0a0f 0%, #12121a 40%, #0d0d12 100%) !important;
|
||||||
}
|
color: #c8c8d0 !important;
|
||||||
.dark-panel {
|
}
|
||||||
border: 1px solid #2a2a35;
|
.dark-panel {
|
||||||
border-radius: 8px;
|
border: 1px solid #2a2a35;
|
||||||
padding: 16px;
|
border-radius: 8px;
|
||||||
background: rgba(18, 18, 26, 0.85);
|
padding: 16px;
|
||||||
margin-bottom: 12px;
|
background: rgba(18, 18, 26, 0.85);
|
||||||
}
|
margin-bottom: 12px;
|
||||||
h1, h2, h3 { color: #e8e8f0 !important; letter-spacing: 0.05em; }
|
}
|
||||||
.status-bar {
|
h1, h2, h3 { color: #e8e8f0 !important; letter-spacing: 0.05em; }
|
||||||
font-family: 'Consolas', 'Monaco', monospace;
|
.status-bar {
|
||||||
font-size: 0.85em;
|
font-family: 'Consolas', 'Monaco', monospace;
|
||||||
color: #7a7a90;
|
font-size: 0.85em;
|
||||||
}
|
color: #7a7a90;
|
||||||
footer { visibility: hidden; }
|
}
|
||||||
"""
|
footer { visibility: hidden; }
|
||||||
|
"""
|
||||||
|
|
||||||
def build_app() -> gr.Blocks:
|
|
||||||
"""构建 Gradio 主界面。"""
|
def build_theme() -> gr.themes.Base:
|
||||||
theme = gr.themes.Monochrome(
|
"""构建暗黑科技风 Monochrome 主题（Gradio 6.0 需在 launch() 传入）。"""
|
||||||
primary_hue="slate",
|
return gr.themes.Monochrome(
|
||||||
secondary_hue="gray",
|
primary_hue="slate",
|
||||||
neutral_hue="slate",
|
secondary_hue="gray",
|
||||||
font=[gr.themes.GoogleFont("Inter"), "system-ui", "sans-serif"],
|
neutral_hue="slate",
|
||||||
font_mono=[gr.themes.GoogleFont("JetBrains Mono"), "Consolas", "monospace"],
|
font=[gr.themes.GoogleFont("Inter"), "system-ui", "sans-serif"],
|
||||||
).set(
|
font_mono=[gr.themes.GoogleFont("JetBrains Mono"), "Consolas", "monospace"],
|
||||||
body_background_fill="#0a0a0f",
|
).set(
|
||||||
body_background_fill_dark="#0a0a0f",
|
body_background_fill="#0a0a0f",
|
||||||
block_background_fill="#12121a",
|
body_background_fill_dark="#0a0a0f",
|
||||||
block_background_fill_dark="#12121a",
|
block_background_fill="#12121a",
|
||||||
block_border_color="#2a2a35",
|
block_background_fill_dark="#12121a",
|
||||||
block_label_text_color="#9090a0",
|
block_border_color="#2a2a35",
|
||||||
input_background_fill="#1a1a24",
|
block_label_text_color="#9090a0",
|
||||||
button_primary_background_fill="#3a3a50",
|
input_background_fill="#1a1a24",
|
||||||
button_primary_background_fill_hover="#4a4a60",
|
button_primary_background_fill="#3a3a50",
|
||||||
)
|
button_primary_background_fill_hover="#4a4a60",
|
||||||
|
)
|
||||||
with gr.Blocks(
|
|
||||||
title="Trading Studio | 交易复盘配音中控",
|
|
||||||
theme=theme,
|
def build_app() -> gr.Blocks:
|
||||||
css=CUSTOM_CSS,
|
"""构建 Gradio 主界面。"""
|
||||||
) as demo:
|
with gr.Blocks(
|
||||||
gr.Markdown(
|
title="Trading Studio | 交易复盘配音中控",
|
||||||
f"""
|
) as demo:
|
||||||
# ⚡ Trading Studio
|
gr.Markdown(
|
||||||
**本地量化交易复盘 → B 站配音生产流水线**
|
f"""
|
||||||
|
# ⚡ Trading Studio
|
||||||
`Whisper(CUDA)` → `Gemma4 @ {OLLAMA_URL}` → `ChatTTS(CUDA)`
|
**本地量化交易复盘 → B 站配音生产流水线**
|
||||||
|
|
||||||
> 仓库: [{GIT_REPO_URL}]({GIT_REPO_URL})
|
`Whisper(CUDA)` → `Gemma4 @ {OLLAMA_URL}` → `ChatTTS(CUDA)`
|
||||||
""",
|
|
||||||
elem_classes=["dark-panel"],
|
> 仓库: [{GIT_REPO_URL}]({GIT_REPO_URL})
|
||||||
)
|
""",
|
||||||
|
elem_classes=["dark-panel"],
|
||||||
with gr.Row():
|
)
|
||||||
ollama_status = gr.Textbox(
|
|
||||||
label="Ollama 节点",
|
with gr.Row():
|
||||||
value=f"模型: {MODEL_NAME}",
|
ollama_status = gr.Textbox(
|
||||||
interactive=False,
|
label="Ollama 节点",
|
||||||
scale=3,
|
value="检测中...",
|
||||||
elem_classes=["status-bar"],
|
interactive=False,
|
||||||
)
|
scale=3,
|
||||||
speaker_status = gr.Textbox(
|
elem_classes=["status-bar"],
|
||||||
label="音色状态",
|
)
|
||||||
value="检测中...",
|
speaker_status = gr.Textbox(
|
||||||
interactive=False,
|
label="音色状态",
|
||||||
scale=2,
|
value="检测中...",
|
||||||
elem_classes=["status-bar"],
|
interactive=False,
|
||||||
)
|
scale=2,
|
||||||
refresh_btn = gr.Button("🔄 刷新状态", scale=1)
|
elem_classes=["status-bar"],
|
||||||
|
)
|
||||||
refresh_btn.click(
|
refresh_btn = gr.Button("🔄 刷新状态", scale=1)
|
||||||
fn=lambda: (ui_check_ollama(), ui_speaker_status()),
|
|
||||||
outputs=[ollama_status, speaker_status],
|
refresh_btn.click(
|
||||||
)
|
fn=lambda: (ui_check_ollama(), ui_speaker_status()),
|
||||||
|
outputs=[ollama_status, speaker_status],
|
||||||
with gr.Tabs():
|
)
|
||||||
# ---- Tab 1: 音色锁定 ----
|
|
||||||
with gr.Tab("🎙️ 音色锁定"):
|
with gr.Tabs():
|
||||||
gr.Markdown(
|
# ---- Tab 1: 音色锁定 ----
|
||||||
"上传 **10-30 秒** 干净人声样本，系统将提取 Speaker Embedding "
|
with gr.Tab("🎙️ 音色锁定"):
|
||||||
f"并保存至 `{SPEAKER_EMB_PATH.name}`，后续合成 100% 还原音色。"
|
gr.Markdown(
|
||||||
)
|
"上传 **10-30 秒** 干净人声样本，系统将提取 Speaker Embedding "
|
||||||
with gr.Row():
|
f"并保存至 `{SPEAKER_EMB_PATH.name}`，后续合成 100% 还原音色。"
|
||||||
spk_audio = gr.Audio(
|
)
|
||||||
label="参考人声（碎碎念盲录样本）",
|
with gr.Row():
|
||||||
type="filepath",
|
spk_audio = gr.Audio(
|
||||||
sources=["upload", "microphone"],
|
label="参考人声（碎碎念盲录样本）",
|
||||||
)
|
type="filepath",
|
||||||
spk_transcript = gr.Textbox(
|
sources=["upload", "microphone"],
|
||||||
label="参考音频精确转写（可选，提升还原度）",
|
)
|
||||||
placeholder="尽量与参考音频内容完全一致...",
|
spk_transcript = gr.Textbox(
|
||||||
lines=6,
|
label="参考音频精确转写（可选，提升还原度）",
|
||||||
)
|
placeholder="尽量与参考音频内容完全一致...",
|
||||||
lock_btn = gr.Button("🔒 锁定音色", variant="primary")
|
lines=6,
|
||||||
lock_log = gr.Textbox(label="锁定结果", lines=4, interactive=False)
|
)
|
||||||
lock_btn.click(ui_lock_speaker, [spk_audio, spk_transcript], lock_log)
|
lock_btn = gr.Button("🔒 锁定音色", variant="primary")
|
||||||
|
lock_log = gr.Textbox(label="锁定结果", lines=4, interactive=False)
|
||||||
# ---- Tab 2: 分步操作 ----
|
lock_btn.click(
|
||||||
with gr.Tab("🔧 分步流水线"):
|
ui_lock_speaker,
|
||||||
with gr.Row():
|
[spk_audio, spk_transcript],
|
||||||
with gr.Column(scale=1):
|
[lock_log, speaker_status],
|
||||||
gr.Markdown("### Step 1 · 音频极速识别")
|
)
|
||||||
rec_audio = gr.Audio(
|
|
||||||
label="交易复盘碎碎念录音",
|
# ---- Tab 2: 分步操作 ----
|
||||||
type="filepath",
|
with gr.Tab("🔧 分步流水线"):
|
||||||
sources=["upload", "microphone"],
|
with gr.Row():
|
||||||
)
|
with gr.Column(scale=1):
|
||||||
transcribe_btn = gr.Button("⚡ Faster-Whisper 识别", variant="primary")
|
gr.Markdown("### Step 1 · 音频极速识别")
|
||||||
transcribe_log = gr.Textbox(label="识别日志", lines=2, interactive=False)
|
rec_audio = gr.Audio(
|
||||||
|
label="交易复盘碎碎念录音",
|
||||||
with gr.Column(scale=1):
|
type="filepath",
|
||||||
gr.Markdown("### Step 2 · Gemma4 纪律审判")
|
sources=["upload", "microphone"],
|
||||||
raw_text = gr.Textbox(
|
)
|
||||||
label="转写原文（可编辑）",
|
transcribe_btn = gr.Button("⚡ Faster-Whisper 识别", variant="primary")
|
||||||
lines=10,
|
transcribe_log = gr.Textbox(label="识别日志", lines=2, interactive=False)
|
||||||
placeholder="识别结果将显示在此，也可手动粘贴...",
|
|
||||||
)
|
with gr.Column(scale=1):
|
||||||
polish_btn = gr.Button("⚖️ 远程 Gemma4 严厉润色", variant="primary")
|
gr.Markdown("### Step 2 · Gemma4 纪律审判")
|
||||||
polish_log = gr.Textbox(label="润色日志", lines=2, interactive=False)
|
raw_text = gr.Textbox(
|
||||||
|
label="转写原文（可编辑）",
|
||||||
with gr.Column(scale=1):
|
lines=10,
|
||||||
gr.Markdown("### Step 3 · ChatTTS 配音合成")
|
placeholder="识别结果将显示在此，也可手动粘贴...",
|
||||||
polished_text = gr.Textbox(
|
)
|
||||||
label="润色配音稿（可编辑）",
|
polish_btn = gr.Button("⚖️ 远程 Gemma4 严厉润色", variant="primary")
|
||||||
lines=10,
|
polish_log = gr.Textbox(label="润色日志", lines=2, interactive=False)
|
||||||
placeholder="润色结果将显示在此...",
|
|
||||||
)
|
with gr.Column(scale=1):
|
||||||
synth_btn = gr.Button("🔊 合成配音 WAV", variant="primary")
|
gr.Markdown("### Step 3 · ChatTTS 配音合成")
|
||||||
synth_log = gr.Textbox(label="合成日志", lines=2, interactive=False)
|
polished_text = gr.Textbox(
|
||||||
output_audio = gr.Audio(label="成品配音", type="filepath")
|
label="润色配音稿（可编辑）",
|
||||||
|
lines=10,
|
||||||
transcribe_btn.click(ui_transcribe, rec_audio, [raw_text, transcribe_log])
|
placeholder="润色结果将显示在此...",
|
||||||
polish_btn.click(ui_polish, raw_text, [polished_text, polish_log])
|
)
|
||||||
synth_btn.click(ui_synthesize, polished_text, [output_audio, synth_log])
|
synth_btn = gr.Button("🔊 合成配音 WAV", variant="primary")
|
||||||
|
synth_log = gr.Textbox(label="合成日志", lines=2, interactive=False)
|
||||||
# ---- Tab 3: 一键生产 ----
|
output_audio = gr.Audio(label="成品配音", type="filepath")
|
||||||
with gr.Tab("🚀 一键生产"):
|
|
||||||
gr.Markdown(
|
transcribe_btn.click(ui_transcribe, rec_audio, [raw_text, transcribe_log])
|
||||||
"上传碎碎念录音，系统自动完成 **识别 → 润色 → 合成** 全流程。"
|
polish_btn.click(ui_polish, raw_text, [polished_text, polish_log])
|
||||||
)
|
synth_btn.click(ui_synthesize, polished_text, [output_audio, synth_log])
|
||||||
with gr.Row():
|
|
||||||
pipe_audio = gr.Audio(
|
# ---- Tab 3: 一键生产 ----
|
||||||
label="复盘录音",
|
with gr.Tab("🚀 一键生产"):
|
||||||
type="filepath",
|
gr.Markdown(
|
||||||
sources=["upload", "microphone"],
|
"上传碎碎念录音，系统自动完成 **识别 → 润色 → 合成** 全流程。"
|
||||||
)
|
)
|
||||||
pipe_manual = gr.Textbox(
|
with gr.Row():
|
||||||
label="或手动输入转写（跳过识别）",
|
pipe_audio = gr.Audio(
|
||||||
lines=4,
|
label="复盘录音",
|
||||||
placeholder="若已有转写文本，可直接粘贴，留空则走 Whisper 识别",
|
type="filepath",
|
||||||
)
|
sources=["upload", "microphone"],
|
||||||
skip_polish_cb = gr.Checkbox(
|
)
|
||||||
label="跳过 Gemma4 润色（仅测试 TTS）",
|
pipe_manual = gr.Textbox(
|
||||||
value=False,
|
label="或手动输入转写（跳过识别）",
|
||||||
)
|
lines=4,
|
||||||
pipeline_btn = gr.Button("▶ 启动全流程", variant="primary", size="lg")
|
placeholder="若已有转写文本，可直接粘贴，留空则走 Whisper 识别",
|
||||||
pipeline_log = gr.Textbox(label="流水线日志", lines=6, interactive=False)
|
)
|
||||||
with gr.Row():
|
skip_polish_cb = gr.Checkbox(
|
||||||
pipe_raw = gr.Textbox(label="转写原文", lines=6)
|
label="跳过 Gemma4 润色（仅测试 TTS）",
|
||||||
pipe_polished = gr.Textbox(label="润色稿", lines=6)
|
value=False,
|
||||||
pipe_output = gr.Audio(label="成品配音", type="filepath")
|
)
|
||||||
|
pipeline_btn = gr.Button("▶ 启动全流程", variant="primary", size="lg")
|
||||||
pipeline_btn.click(
|
pipeline_log = gr.Textbox(label="流水线日志", lines=6, interactive=False)
|
||||||
ui_full_pipeline,
|
with gr.Row():
|
||||||
[pipe_audio, skip_polish_cb, pipe_manual],
|
pipe_raw = gr.Textbox(label="转写原文", lines=6)
|
||||||
[pipe_raw, pipe_polished, pipe_output, pipeline_log],
|
pipe_polished = gr.Textbox(label="润色稿", lines=6)
|
||||||
)
|
pipe_output = gr.Audio(label="成品配音", type="filepath")
|
||||||
|
|
||||||
demo.load(
|
pipeline_btn.click(
|
||||||
fn=lambda: (ui_check_ollama(), ui_speaker_status()),
|
ui_full_pipeline,
|
||||||
outputs=[ollama_status, speaker_status],
|
[pipe_audio, skip_polish_cb, pipe_manual],
|
||||||
)
|
[pipe_raw, pipe_polished, pipe_output, pipeline_log],
|
||||||
|
)
|
||||||
return demo
|
|
||||||
|
demo.load(
|
||||||
|
fn=lambda: (ui_check_ollama(), ui_speaker_status()),
|
||||||
def main() -> None:
|
outputs=[ollama_status, speaker_status],
|
||||||
"""主入口：启动 Gradio 服务。"""
|
)
|
||||||
logger.info("Trading Studio 启动中... HOST=%s PORT=%s", HOST, PORT)
|
|
||||||
app = build_app()
|
return demo
|
||||||
app.launch(
|
|
||||||
server_name=HOST,
|
|
||||||
server_port=PORT,
|
def main() -> None:
|
||||||
share=False,
|
"""主入口：启动 Gradio 服务。"""
|
||||||
show_error=True,
|
logger.info("Trading Studio 启动中... HOST=%s PORT=%s", HOST, PORT)
|
||||||
allowed_paths=[str(Path(__file__).resolve().parent / "outputs")],
|
app = build_app()
|
||||||
)
|
app.launch(
|
||||||
|
server_name=HOST,
|
||||||
|
server_port=PORT,
|
||||||
if __name__ == "__main__":
|
share=False,
|
||||||
main()
|
show_error=True,
|
||||||
|
theme=build_theme(),
|
||||||
|
css=CUSTOM_CSS,
|
||||||
|
allowed_paths=[str(Path(__file__).resolve().parent / "outputs")],
|
||||||
|
)
|
||||||
|
|
||||||
|
|
||||||
|
if __name__ == "__main__":
|
||||||
|
main()
|
||||||
|
|||||||
Reference in New Issue
Block a user