STBack23 commited on
Commit
8fb2330
·
verified ·
1 Parent(s): 5d0d4a3

Update OmniVoice VI voices + Colab tools

Browse files
Files changed (1) hide show
  1. colab/Omivoice_VI_Colab.ipynb +45 -15
colab/Omivoice_VI_Colab.ipynb CHANGED
@@ -6,13 +6,13 @@
6
  "source": [
7
  "# OmniVoice — Lồng tiếng tiếng Việt trên Colab\n",
8
  "\n",
9
- "[![Open In Colab](https://colab.research.google.com/assets/colab-badge.svg)](https://colab.research.google.com/github/YOUR_USER/Omivoice/blob/main/hf/colab/Omivoice_VI_Colab.ipynb)\n",
10
  "\n",
11
- "Notebook chạy [OmniVoice](https://huggingface.co/k2-fsa/OmniVoice) với **6 giọng Việt** và pipeline **SRT lồng tiếng** (`native` — không cắt ngắt, tràn vào khoảng trống sau cue).\n",
 
 
12
  "\n",
13
- "**Yêu cầu:** Runtime **GPU** (T4+).\n",
14
- "\n",
15
- "**Trước khi chạy:** sửa `HF_REPO` thành dataset Hugging Face bạn đã upload."
16
  ],
17
  "id": "029da2d4"
18
  },
@@ -28,9 +28,13 @@
28
  "cell_type": "code",
29
  "metadata": {},
30
  "source": [
31
- "# Repo dataset trên HF — xem hf/README.md\n",
32
  "HF_REPO = \"STBack23/omnivoice-vi\"\n",
33
  "\n",
 
 
 
 
34
  "DEFAULT_VOICE = \"ngoc_huyen\"\n",
35
  "MERGE_MODE = \"native\" # native | cascade | fit | strict\n",
36
  "NUM_STEP = 32"
@@ -96,9 +100,9 @@
96
  " login(\"hf_xxxxxxxx\") # thay token của bạn\n",
97
  " print(\"Đã login HF (token)\")"
98
  ],
99
- "id": "e659ec6b",
100
  "execution_count": null,
101
- "outputs": []
 
102
  },
103
  {
104
  "cell_type": "markdown",
@@ -224,12 +228,32 @@
224
  "cell_type": "code",
225
  "metadata": {},
226
  "source": [
227
- "from google.colab import files\n",
 
 
228
  "\n",
229
- "print(\"Upload file .srt:\")\n",
230
- "uploaded = files.upload()\n",
231
- "SRT_PATH = Path(\"/content\") / list(uploaded.keys())[0]\n",
232
- "print(f\"File: {SRT_PATH}\")"
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
233
  ],
234
  "execution_count": null,
235
  "outputs": [],
@@ -244,7 +268,9 @@
244
  "result = speak.run_srt_pipeline(\n",
245
  " profile_path,\n",
246
  " SRT_PATH,\n",
 
247
  " merge=True,\n",
 
248
  " merge_mode=MERGE_MODE,\n",
249
  " num_step=NUM_STEP,\n",
250
  " model=MODEL,\n",
@@ -254,8 +280,12 @@
254
  "print(result[\"log\"])\n",
255
  "merged = result.get(\"merged_wav\")\n",
256
  "if merged:\n",
257
- " display(Audio(merged))\n",
258
- " files.download(merged)"
 
 
 
 
259
  ],
260
  "execution_count": null,
261
  "outputs": [],
 
6
  "source": [
7
  "# OmniVoice — Lồng tiếng tiếng Việt trên Colab\n",
8
  "\n",
9
+ "Notebook chạy [OmniVoice](https://huggingface.co/k2-fsa/OmniVoice) với **6 giọng Việt** và pipeline **SRT lồng tiếng** (`native`).\n",
10
  "\n",
11
+ "**Luồng Drive (mặc định):**\n",
12
+ "1. Đặt file `.srt` vào **`MyDrive/Omivoice/`** (tên file cấu hình ở cell 0)\n",
13
+ "2. Chạy notebook → kết quả ghi lại **`MyDrive/Omivoice/`** (`*_merged.wav`, `output/...`)\n",
14
  "\n",
15
+ "**Yêu cầu:** Runtime **GPU** (T4+)."
 
 
16
  ],
17
  "id": "029da2d4"
18
  },
 
28
  "cell_type": "code",
29
  "metadata": {},
30
  "source": [
31
+ "# Hugging Face (giọng + speak.py)\n",
32
  "HF_REPO = \"STBack23/omnivoice-vi\"\n",
33
  "\n",
34
+ "# Google Drive — MyDrive/Omivoice/\n",
35
+ "DRIVE_FOLDER = \"Omivoice\"\n",
36
+ "SRT_FILE = \"input.srt\" # file SRT trong thư mục Omivoice (có thể \"video\" hoặc \"video.srt\")\n",
37
+ "\n",
38
  "DEFAULT_VOICE = \"ngoc_huyen\"\n",
39
  "MERGE_MODE = \"native\" # native | cascade | fit | strict\n",
40
  "NUM_STEP = 32"
 
100
  " login(\"hf_xxxxxxxx\") # thay token của bạn\n",
101
  " print(\"Đã login HF (token)\")"
102
  ],
 
103
  "execution_count": null,
104
+ "outputs": [],
105
+ "id": "e659ec6b"
106
  },
107
  {
108
  "cell_type": "markdown",
 
228
  "cell_type": "code",
229
  "metadata": {},
230
  "source": [
231
+ "from google.colab import drive\n",
232
+ "\n",
233
+ "drive.mount(\"/content/drive\")\n",
234
  "\n",
235
+ "DRIVE_OMIVOICE = Path(\"/content/drive/MyDrive\") / DRIVE_FOLDER\n",
236
+ "DRIVE_OMIVOICE.mkdir(parents=True, exist_ok=True)\n",
237
+ "\n",
238
+ "srt_name = SRT_FILE if SRT_FILE.lower().endswith(\".srt\") else f\"{SRT_FILE}.srt\"\n",
239
+ "SRT_PATH = DRIVE_OMIVOICE / srt_name\n",
240
+ "\n",
241
+ "if not SRT_PATH.exists():\n",
242
+ " existing = sorted(DRIVE_OMIVOICE.glob(\"*.srt\"))\n",
243
+ " hint = \"\\n \".join(str(p.name) for p in existing) if existing else \"(chưa có file .srt)\"\n",
244
+ " raise FileNotFoundError(\n",
245
+ " f\"Không thấy {SRT_PATH}\\n\"\n",
246
+ " f\"Đặt file vào MyDrive/{DRIVE_FOLDER}/ hoặc sửa SRT_FILE.\\n\"\n",
247
+ " f\"Các file .srt hiện có:\\n {hint}\"\n",
248
+ " )\n",
249
+ "\n",
250
+ "MERGE_OUT = DRIVE_OMIVOICE / f\"{SRT_PATH.stem}_merged.wav\"\n",
251
+ "OUTPUT_DIR = DRIVE_OMIVOICE / \"output\" / PROFILE[\"slug\"] / SRT_PATH.stem\n",
252
+ "\n",
253
+ "print(f\"Drive: {DRIVE_OMIVOICE}\")\n",
254
+ "print(f\"Input: {SRT_PATH}\")\n",
255
+ "print(f\"Output: {MERGE_OUT}\")\n",
256
+ "print(f\"Cues: {OUTPUT_DIR}\")"
257
  ],
258
  "execution_count": null,
259
  "outputs": [],
 
268
  "result = speak.run_srt_pipeline(\n",
269
  " profile_path,\n",
270
  " SRT_PATH,\n",
271
+ " output_dir=OUTPUT_DIR,\n",
272
  " merge=True,\n",
273
+ " merge_output=MERGE_OUT,\n",
274
  " merge_mode=MERGE_MODE,\n",
275
  " num_step=NUM_STEP,\n",
276
  " model=MODEL,\n",
 
280
  "print(result[\"log\"])\n",
281
  "merged = result.get(\"merged_wav\")\n",
282
  "if merged:\n",
283
+ " display(Audio(str(merged)))\n",
284
+ "\n",
285
+ "print(\"\\n--- Đã lưu trên Drive ---\")\n",
286
+ "print(f\"WAV gộp: {result.get('merged_wav')}\")\n",
287
+ "print(f\"Từng cue: {result.get('output_dir')}\")\n",
288
+ "print(f\"SRT shifted: {result.get('shifted_srt')}\")"
289
  ],
290
  "execution_count": null,
291
  "outputs": [],