Audio Generation

Audio Generation - AI Speech, Voiceover, and Sound Powered by CellCog

Create professional audio with AI — voiceovers, music, sound effects, and personalized avatar voices.

CellCog provides three voice providers, each with different strengths. Choose based on your needs:

How to Use

For your first CellCog task in a session, read the cellcog skill for the full SDK reference — file handling, chat modes, timeouts, and more.

OpenClaw (fire-and-forget):

result = client.create_chat(
    prompt="[your task prompt]",
    notify_session_key="agent:main:main",
    task_label="my-task",
    chat_mode="agent",
)

All agents except OpenClaw (blocks until done):

from cellcog import CellCogClient
client = CellCogClient(agent_provider="openclaw|cursor|claude-code|codex|...")
result = client.create_chat(
    prompt="[your task prompt]",
    task_label="my-task",
    chat_mode="agent",
)
print(result["message"])

---

Voice Providers

OpenAI (Default)

Best for standard narration, voiceovers, and single-speaker content with precise delivery control.

Key strength: Natural-language style instructions — describe the accent, tone, pacing, and emotion you want.

8 built-in voices:

Best quality: cedar (male), marin (female).

Style customization examples:

• "Warm conversational tone, medium pace, slight enthusiasm when mentioning features. American accent."
• "Deep, hushed, enigmatic, with a slow deliberate cadence — true crime narrator style."
• "Heavy French accent, sophisticated yet friendly, moderate pacing with deliberate pauses."

---

ElevenLabs

Best for emotional delivery, dramatic content, character voices, and audiobook narration.

Key strength: Emotion tags embedded directly in text — [laughs], [sighs], [whispers], [excited], [sarcastic]. Plus 100+ diverse pre-made voices.

Emotion tags (use sparingly — 1-2 per paragraph):

Example prompt:

"Generate speech using ElevenLabs with a warm British male voice: 'And then, just when everyone thought it was over... [pause] [whispers] it wasn't.'"

---

MiniMax

Best for cloned voices (avatars) and fine-grained voice control.

Key strength: MiniMax Speech 2.8 HD — studio-grade audio quality. Supports avatar cloned voice IDs for personalized content, plus 17+ standard pre-made voices with granular speed, pitch, and volume control.

Standard voices include: Deep_Voice_Man, Calm_Woman, Casual_Guy, Lively_Girl, Wise_Woman, Friendly_Person, Young_Knight, Elegant_Man, and more.

Voice settings: emotion (happy/sad/angry/neutral/etc.), speed (0.5–2.0), volume (0–10), pitch (-12 to 12).

---

Avatar / Cloned Voice

Users can create avatars on CellCog with their own cloned voice. When an avatar has a cloned voice, CellCog uses the MiniMax provider to generate speech that sounds like that person.

How it works:

• The user creates an avatar on cellcog.ai and uploads voice samples
• CellCog clones their voice using MiniMax Speech 2.8 HD
• Any audio request referencing that avatar uses their cloned voice

Example prompt:

"Generate a voiceover using my avatar Luna's voice: 'Welcome to our quarterly update. I'm excited to share some incredible results with you today.'"

This is powerful for creating consistent, personalized content — marketing videos, podcast intros, course narration — all in the user's own voice.

---

Sound Effects (SFX)

CellCog generates standalone sound effects from text descriptions. Royalty-free, 0.1 to 30 seconds.

Example prompts:

• "Generate a sound effect of heavy rain hitting a metal roof with occasional thunder, 10 seconds"
• "Create a crispy footsteps-on-fresh-snow sound effect, 5 seconds"
• "Generate an echoing door slam in a large empty warehouse"

Tips for better SFX:

• Be specific about textures and environment
• Specify duration when exact length matters
• For ambient audio longer than 30 seconds, generate a short loopable segment and extend with ffmpeg

---

Music Generation

Create original music from text descriptions. 3 seconds to 10 minutes. Royalty-free.

Capabilities:

• Any genre or genre fusion
• Instrumental and vocal tracks (specify if you want vocals)
• Complex arrangements, mood transitions, and energy dynamics
• Describe what you want — the model handles music theory

Example prompts:

• "Create 2 minutes of calm lo-fi hip-hop background music with soft piano and mellow beats, 75 BPM"
• "Generate a 15-second upbeat tech podcast intro jingle"
• "Create 90 seconds of cinematic orchestral music — start soft and inspiring, build to a confident crescendo"
• "Generate a 3-minute pop song about summer adventures with female vocals"

For precise section-by-section control (exact timing per section), describe your composition plan in detail — CellCog handles the structure.

All generated music is royalty-free — use commercially without attribution or licensing fees.

---

Multi-Language Support

All three voice providers support 40+ languages. Provide speech text in the target language:

English, Spanish, French, German, Italian, Portuguese, Chinese (Mandarin/Cantonese), Japanese, Korean, Hindi, Arabic, Russian, Polish, Dutch, Turkish, and many more.

---

Chat Mode

Use chat_mode="agent" for all audio tasks. Audio generation executes efficiently in agent mode — no need for agent team.

---

Tips for Better Audio

1. Choose the right provider: OpenAI for standard narration, ElevenLabs for emotional/dramatic, MiniMax for cloned voices
2. Provide the complete script: Write out exactly what should be spoken — don't say "something about our product"
3. Include style instructions: "Confident but warm", "slow and deliberate", "with slight excitement"
4. For music: Specify duration, mood, genre, and tempo (BPM if you know it)
5. Pronunciation guidance: For names or technical terms, add hints: "CellCog (pronounced SELL-kog)"
6. For ElevenLabs emotion tags: Use sparingly — 1-2 per paragraph. Tags affect all subsequent text until a new tag.

---

If CellCog is not installed

Claude Code, Cursor, Codex + 70 more agents: npx skills add cellcog/skills --skill cellcog OpenClaw: clawhub install cellcog CellCog plugin users: run /cellcog-setup (or /cellcog:cellcog-setup depending on your tool) Manual setup: pip install -U cellcog and set CELLCOG_API_KEY. See the cellcog skill for SDK reference.