Audio AI cho personal brand — voice clone (ElevenLabs/HeyGen Voice/Vbee), podcast workflow, audiobook, voiceover video. 3 use case: voiceover ngan TikTok/Reels (energetic), podcast 30-60 phut (conversational), audiobook (mid-tempo). Repurpose 1:10 (1 podcast → 10 short clip). Trigger: 'voice clone', 'ElevenLabs giong noi', 'podcast AI', 'audio AI', 'Descript voice', 'Riverside podcast', 'voiceover AI'.
npx skills add https://github.com/minhnv0807/ai-business-skills --skill 25-voice-clone-podcast
> Skill nay tap trung vao audio AI — voice clone, podcast, audiobook, voiceover.
> Bo sung cho 24-ai-avatar-production (video) — ket hop ca 2 de phu het content stack.
Audio AI la cong nghe tao ra giong noi nhan tao gan giong nguoi that — tu sample
giong cua ban, AI hoc va tao ra giong nhan ban (voice clone). Ban viet text →
AI doc thay (Text-to-Speech).
Khac biet voi video AI:
| Tinh huong | Chon audio AI | Chon video AI |
|-----------|---------------|---------------|
| Noi dung dai (>10 phut) | YES — podcast format | NO — too long for video |
| Khong muon len hinh | YES | NO |
| Can tao volume content nhanh | YES — 1 podcast = 10 short | YES nhung ton hon |
| Audience nghe khi lai xe / tap gym | YES | NO |
| Can visual de demo | NO | YES |
| Personal brand thought leader | YES — podcast = authority | YES — neu da co face brand |
| Cong viec | Thoi gian | Chi phi (USD/thang) |
|-----------|-----------|----------------------|
| Voice clone setup | 30-60 phut | $5-22 (ElevenLabs Starter/Pro) |
| Voiceover 60s (TikTok) | 5-10 phut | $5-22 |
| Podcast 30 phut (solo) | 1-2 gio | $22-99 (ElevenLabs + Riverside) |
| Audiobook 1 chuong (15 phut) | 30-45 phut | $22-99 |
| Repurpose 1 podcast → 10 clip | 1-2 gio | $0-30 (Descript/Opus) |
Hoi toi da 4 cau truoc khi bat dau:
> Dua tren 4 cau tra loi, chon use case + tool stack phu hop.
| Tieu chi | Yeu cau toi thieu | Toi uu |
|----------|---------------------|--------|
| Thoi luong | 1 phut (Free tier) | 3-5 phut (Pro tier) |
| Phong | Yen tinh, khong vang | Treo chan, rem, sach hap thu am |
| Mic | iPhone + tai nghe co mic | Condenser mic (AT2020, $80-100) |
| Distance | 20-30cm | 15-20cm voi pop filter |
| Format | MP3 128kbps | WAV 44.1kHz |
| Noi dung | 1 doan van da chuan bi | 3 doan van: business / casual / emotional |
> Reference day du: references/voice-clone-prompts-vn.md — 3 sample script
> theo vung giong (Bac/Trung/Nam) va 3 topic (business/lifestyle/educational).
| Tool | VN voice clone | Gia/thang | Setup time | Best for |
|------|----------------|-----------|------------|----------|
| ElevenLabs Pro | Tot (8/10) | $22 | 30 phut | Multi-language, content creator |
| HeyGen Voice | Trung binh (6/10) | Bundle voi avatar | 15 phut | Combo voi video AI |
| Vbee Pro | Xuat sac (9.5/10) | 199K-499K VND | 45 phut | VN-only, broadcast TTS |
| Descript Overdub | Trung binh (6/10) | $24 (Hobbyist) | 30 phut | Podcast editing |
| Resemble.ai | Trung binh (7/10) | $30 | 1 gio | API integration, custom |
Khuyen nghi:
THOA THUAN SU DUNG VOICE CLONE
Toi, [Ho ten], CMND/CCCD: [so], dong y cho [Brand/Cong ty]:
1. Su dung sample giong noi cua toi de tao voice clone AI
2. Su dung voice clone trong [pham vi: noi bo / quang cao / podcast / etc.]
3. Thoi han: [tu DD/MM/YYYY den DD/MM/YYYY]
4. Quyen rut lai: Toi co quyen yeu cau xoa voice clone bat ky luc nao
bang van ban, brand co 7 ngay de xoa hoan toan.
5. Cong khai: Brand cam ket disclose "AI voice" theo quy tac VN.
Ky ten: ____________ Ngay: ____________
Spec:
Script template (30s):
[HOOK 0-3s] "Ban co biet [stat shocking]?"
[PROBLEM 3-10s] "Hau het moi nguoi van dang [vong xoay sai]"
[SOLUTION 10-22s] "Toi da thu [phuong phap], va day la 3 dieu..."
[PAYOFF 22-27s] "Ket qua: [so cu the]"
[CTA 27-30s] "Comment 'YES' de minh gui chi tiet"
Voice settings (ElevenLabs):
Cau truc:
Pacing:
Sound design:
Voice settings (ElevenLabs):
Cau truc:
Pacing:
Consistency check (quan trong nhat):
Voice settings (Vbee, neu VN):
| Tool | Gia/thang | VN voice native | EN voice | Setup | Pros | Cons | Best for |
|------|-----------|-----------------|----------|-------|------|------|----------|
| ElevenLabs | $5-99 | 8/10 | 10/10 | 30 phut | Multi-lang, voice clone tot | VN phat am vai tu kho | Multi-lang creator |
| Vbee | 199K-1.5M VND | 9.5/10 | 6/10 | 45 phut | VN tot nhat, da giong vung | Khong manh EN | VN-only audio |
| HeyGen Voice | Bundle voi avatar | 6/10 | 8/10 | 15 phut | Combo voi avatar | Voice clone don dieu | Combo voi video |
| Descript | $24-30 | 6/10 | 9/10 | 30 phut | Audio editing manh | VN voice yeu | Podcast editing |
| Riverside | $19-29 | n/a (recording) | n/a | 5 phut | Studio quality recording | Khong phai TTS | Live podcast |
| Murf | $29-79 | 7/10 | 9/10 | 30 phut | 120+ voice library | Voice clone gioi han | Corporate voiceover |
| PlayHT | $39-99 | 7/10 | 9.5/10 | 30 phut | API tot, instant clone | UI kho | Developer/API |
| Resemble.ai | $30-99 | 7/10 | 9/10 | 1 gio | Custom emotion control | Hoc cao | Brand custom voice |
Combo khuyen nghi 2025-2026:
> Use case: Solo podcaster muon co conversation, khong tim duoc co-host that.
> AI co-host = giong AI thu 2 dong vai dong host, hoi cau + ban tra loi.
Buoc 1: Dinh nghia personality cua AI co-host
Ten: [Ten AI co-host]
Tinh cach: Tro mo, hay hoi sau, doi khi hai huoc nhe
Vai tro: Dat cau hoi cho host, khong tu noi qua nhieu
Phong cach noi: Casual, tu nhien, dung "minh/ban" (khong "toi/anh")
Cap do kien thuc: Trung binh — dat cau hoi nhu listener
Cau cam thuong dung: "Wow, hay quoc!", "Vay nghia la sao?", "Cu the hon nha?"
Buoc 2: Tao voice clone rieng cho AI co-host
Buoc 3: Tool stack
[INTRO]
Host: Chao moi nguoi, hom nay minh va [AI co-host] se ban ve...
AI co-host: Chao cac ban, minh la [ten]. Hom nay minh muon hieu sau ve [topic]
tu goc nhin cua [host]. Bat dau thoi!
[BODY — 5-7 cap Q&A]
AI co-host: [Hoi cau 1 — broad question]
Host: [Tra loi 2-3 phut]
AI co-host: [Hoi follow-up sau hon]
Host: [Tra loi voi vi du cu the]
... lap lai 5-7 lan ...
[OUTRO]
AI co-host: Cam on [host] da chia se. Toi nhat ma minh hoc duoc la...
Host: Cam on [AI co-host]. Cac ban con cau hoi gi, comment ben duoi...
Tip: Viet truoc 7-10 cau hoi cua AI co-host trong document, host tra loi luot.
Sau do generate audio cua AI co-host bang ElevenLabs, ghep vao bang Descript.
[1] Record podcast 60 phut (Riverside)
↓
[2] Transcript tu dong (Descript / Riverside)
↓
[3] Identify hooks (10-15 cau hay)
↓
[4] Cut clips 30-60s moi cau (Opus Clip / Descript)
↓
[5] Add captions (auto-caption)
↓
[6] Distribute ra 4 nen tang
Tim trong transcript nhung cau co dac diem:
> Target: 10-15 hook cho 1 podcast 60 phut. Loc lai 10 clip ngon nhat.
| Nen tang | Format | Thoi luong | Caption | Bonus |
|----------|--------|------------|---------|-------|
| TikTok | 9:16 (1080×1920) | 30-60s | Bold caption tren | Trend audio overlay (volume thap) |
| Instagram Reels | 9:16 | 15-90s | Subtitle dep, font sans-serif | Cover image dep |
| YouTube Shorts | 9:16 | <60s | Auto-caption YouTube | Title chua keyword |
| LinkedIn audio | 1:1 (square video voi audio) | 60-120s | Caption ben duoi | Doc thread bai dai (carousel) |
Pro tip: Moi clip = 1 platform rieng, dung khac caption + cover image. Tang reach.
> Pass: 40+/50 diem. <40 = re-render hoac re-record.
| Tinh huong | Disclosure | Vi tri |
|-----------|-----------|--------|
| Quang cao thuong mai | BAT BUOC | Caption + cuoi audio ("Audio nay su dung voice clone AI") |
| Podcast personal brand | NEN — minh bach | Episode description |
| Audiobook fiction | KHONG bat buoc | Optional — credits cuoi |
| Tin tuc/giao duc | BAT BUOC | Dau audio + caption |
| Noi dung noi bo cong ty | KHONG bat buoc | n/a |
Template disclosure caption:
Audio nay su dung cong nghe voice clone AI
(ElevenLabs / Vbee / [tool ten]). Noi dung do [Ten ban] viet va duyet.
> Reference day du: references/ai-video-disclosure-vn.md — Nghi dinh 147/2024,
> 3 tang disclose, va template cho tung tinh huong (cung ap dung cho audio).
Truoc khi xuat ban audio:
Create beautiful visual art in .png and .pdf documents using design philosophy. You should use this skill when the user asks to create a poster, piece of art, design, or other static piece. Create original visual designs, never copying existing artists' work to avoid copyright violations.
Creating algorithmic art using p5.js with seeded randomness and interactive parameter exploration. Use this when users request creating art using code, generative art, algorithmic art, flow fields, or particle systems. Create original algorithmic art rather than copying existing artists' work to avoid copyright violations.
Improves the quality of images, especially screenshots, by enhancing resolution, sharpness, and clarity. Perfect for preparing images for presentations, documentation, or social media posts.
Downloads videos from YouTube and other platforms for offline viewing, editing, or archival. Handles various formats and quality options.
Lightweight WSI tile extraction and preprocessing. Use for basic slide processing tissue detection, tile extraction, stain normalization for H&E images. Best for simple pipelines, dataset preparation, quick tile-based analysis. For advanced spatial proteomics, multiplexed imaging, or deep learning pipelines use pathml.
Microscopy data management platform. Access images via Python, retrieve datasets, analyze pixels, manage ROIs/annotations, batch processing, for high-content screening and microscopy workflows.
Python library for working with DICOM (Digital Imaging and Communications in Medicine) files. Use this skill when reading, writing, or modifying medical imaging data in DICOM format, extracting pixel data from medical images (CT, MRI, X-ray, ultrasound), anonymizing DICOM files, working with DICOM metadata and tags, converting DICOM images to other formats, handling compressed DICOM data, or processing medical imaging datasets. Applies to tasks involving medical image analysis, PACS systems, radiology workflows, and healthcare imaging applications.
This skill should be used when working with pre-trained transformer models for natural language processing, computer vision, audio, or multimodal tasks. Use for text generation, classification, question answering, translation, summarization, image classification, object detection, speech recognition, and fine-tuning models on custom datasets.
Take minhnv0807/25-voice-clone-podcast from the repository into ~/.claude/skills for personal
use, or into .claude/skills inside a project.
The agent identifies a skill by the name field in its header. Two skills with the
same name cannot sit side by side — one of them will be ignored.