audio-jingle

Name: audio-jingle
Author: mphinance

15stars3forks

How to install audio-jingle?

mphinance/alpha-skills/audio-jingle

$npx -y skills add mphinance/alpha-skills --skill audio-jingle

Installs into the current project.

Summary

Audio generation skill — jingles, beds, voiceover, and sound effects. Routes music requests to Suno V5 / Udio / Lyria, speech to MiniMax TTS / FishAudio / ElevenLabs V3, and SFX to ElevenLabs SFX or AudioCraft. Output is one MP3/WAV file saved to the project folder.

Files · 1

View on GitHub

SKILL.md

1	# Audio Jingle Skill
2
3	Three sub-modes. The active project's `audioKind` decides which one
4	runs:
5
6	\| `audioKind` \| Models we route to \| Plan focus \|
7	\|---\|---\|---\|
8	\| `music` \| Suno V5 (default), Udio, Lyria 2 \| genre + tempo + instrumentation \|
9	\| `speech` \| MiniMax TTS (default), Fish, ElevenLabs V3 \| script + voice + pacing \|
10	\| `sfx` \| ElevenLabs SFX (default), AudioCraft \| texture + impact + duration \|
11
12	## Resource map
13
14	```
15	audio-jingle/
16	├── SKILL.md
17	└── example.html
18	```
19
20	## Workflow
21
22	### Step 0 — Read the project metadata
23
24	`audioKind`, `audioModel`, `audioDuration` (seconds), and (for speech)
25	`voice`. Branch by `audioKind` and use the values verbatim — no
26	clarifying form unless something is marked `(unknown — ask)`.
27
28	Important: `voice` is provider-specific. For `minimax-tts`, `--voice`
29	must be a valid MiniMax `voice_id` (for example `male-qn-qingse`), not
30	a natural-language description. If you only have a prose voice brief
31	("warm female narrator", "neutral Mandarin"), keep that in your plan
32	but omit `--voice` so the daemon's default voice id applies, or ask the
33	user to choose a specific id.
34
35	### Step 1 — Plan
36
37	Music
38	- Genre + reference artists (1-2)
39	- Tempo (BPM) + key
40	- Instrumentation (3-5 instruments max)
41	- Vocals: yes / no / hummed / choir
42	- Mood arc (intro → chorus → outro)
43
44	Speech
45	- Script (final, not draft — TTS runs verbatim)
46	- Voice target + pacing
47	For MiniMax this means a real `voice_id`, not prose in `--voice`
48	- Pronunciation hints for proper nouns / acronyms
49
50	SFX
51	- Texture (impact / whoosh / ambience / foley)
52	- Duration + envelope (sharp attack vs. gentle swell)
53	- Layering note (single hit vs. stacked)
54
55	State the plan in 2-3 sentences before dispatching.
56
57	### Step 2 — Compose the prompt
58
59	Use the format the upstream model prefers. Bind `audioDuration` to the
60	API parameter directly; never put "make it 30 seconds" in prose.
61
62	### Step 3 — Dispatch via the media contract
63
64	Use the unified dispatcher — do not call provider APIs by hand:
65
66	```bash
67	node "$OD_BIN" media generate \
68	--project "$OD_PROJECT_ID" \
69	--surface audio \
70	--audio-kind "<music\|speech\|sfx>" \
71	--model "<audioModel from metadata>" \
72	--duration <audioDuration seconds> \
73	[--voice "<provider voice id (speech only)>"] \
74	--output "<short-slug>-<duration>s.mp3" \
75	--prompt "<assembled prompt from Step 2 — for speech, the literal script>"
76	```
77
78	The command prints one line of JSON: `{"file": {"name": "...", ...}}`.
79	The bytes land in the project; the FileViewer renders the audio
80	transport controls automatically.
81
82	### Step 4 — Hand off
83
84	Reply with: plan summary, the filename returned by the dispatcher, and
85	one sentence on what to try if the user wants a variation (e.g. "swap
86	tempo from 92 to 108 BPM" rather than "make it different").
87
88	## Hard rules
89
90	- TTS runs your script literally. Proof it before dispatching —
91	even one stray comma changes the cadence.
92	- MiniMax TTS rejects free-form voice prose in `--voice`. Use a real
93	MiniMax `voice_id` (for example `male-qn-qingse`) or omit the flag
94	and let the daemon's default voice apply.
95	- Music: under 30s = single section; 30–90s = intro + body; 90s+ =
96	full arc. Don't try to fit a 3-act song into 15 seconds.
97	- SFX: prefer one well-described layer over a paragraph of "make it
98	cool" — generators reward specific texture words.
99	- Save the file every turn. The audio viewer shows transport controls
100	the moment the file lands.

People who liked this also viewed

6 picks

Type

audio-jingle

15stars3forks

View on GitHub

How to install audio-jingle?

mphinance/alpha-skills/audio-jingle

$npx -y skills add mphinance/alpha-skills --skill audio-jingle

Installs into the current project.

Summary

Files · 1

View on GitHub

SKILL.md

1	# Audio Jingle Skill
2
3	Three sub-modes. The active project's `audioKind` decides which one
4	runs:
5
6	\| `audioKind` \| Models we route to \| Plan focus \|
7	\|---\|---\|---\|
8	\| `music` \| Suno V5 (default), Udio, Lyria 2 \| genre + tempo + instrumentation \|
9	\| `speech` \| MiniMax TTS (default), Fish, ElevenLabs V3 \| script + voice + pacing \|
10	\| `sfx` \| ElevenLabs SFX (default), AudioCraft \| texture + impact + duration \|
11
12	## Resource map
13
14	```
15	audio-jingle/
16	├── SKILL.md
17	└── example.html
18	```
19
20	## Workflow
21
22	### Step 0 — Read the project metadata
23
24	`audioKind`, `audioModel`, `audioDuration` (seconds), and (for speech)
25	`voice`. Branch by `audioKind` and use the values verbatim — no
26	clarifying form unless something is marked `(unknown — ask)`.
27
28	Important: `voice` is provider-specific. For `minimax-tts`, `--voice`
29	must be a valid MiniMax `voice_id` (for example `male-qn-qingse`), not
30	a natural-language description. If you only have a prose voice brief
31	("warm female narrator", "neutral Mandarin"), keep that in your plan
32	but omit `--voice` so the daemon's default voice id applies, or ask the
33	user to choose a specific id.
34
35	### Step 1 — Plan
36
37	Music
38	- Genre + reference artists (1-2)
39	- Tempo (BPM) + key
40	- Instrumentation (3-5 instruments max)
41	- Vocals: yes / no / hummed / choir
42	- Mood arc (intro → chorus → outro)
43
44	Speech
45	- Script (final, not draft — TTS runs verbatim)
46	- Voice target + pacing
47	For MiniMax this means a real `voice_id`, not prose in `--voice`
48	- Pronunciation hints for proper nouns / acronyms
49
50	SFX
51	- Texture (impact / whoosh / ambience / foley)
52	- Duration + envelope (sharp attack vs. gentle swell)
53	- Layering note (single hit vs. stacked)
54
55	State the plan in 2-3 sentences before dispatching.
56
57	### Step 2 — Compose the prompt
58
59	Use the format the upstream model prefers. Bind `audioDuration` to the
60	API parameter directly; never put "make it 30 seconds" in prose.
61
62	### Step 3 — Dispatch via the media contract
63
64	Use the unified dispatcher — do not call provider APIs by hand:
65
66	```bash
67	node "$OD_BIN" media generate \
68	--project "$OD_PROJECT_ID" \
69	--surface audio \
70	--audio-kind "<music\|speech\|sfx>" \
71	--model "<audioModel from metadata>" \
72	--duration <audioDuration seconds> \
73	[--voice "<provider voice id (speech only)>"] \
74	--output "<short-slug>-<duration>s.mp3" \
75	--prompt "<assembled prompt from Step 2 — for speech, the literal script>"
76	```
77
78	The command prints one line of JSON: `{"file": {"name": "...", ...}}`.
79	The bytes land in the project; the FileViewer renders the audio
80	transport controls automatically.
81
82	### Step 4 — Hand off
83
84	Reply with: plan summary, the filename returned by the dispatcher, and
85	one sentence on what to try if the user wants a variation (e.g. "swap
86	tempo from 92 to 108 BPM" rather than "make it different").
87
88	## Hard rules
89
90	- TTS runs your script literally. Proof it before dispatching —
91	even one stray comma changes the cadence.
92	- MiniMax TTS rejects free-form voice prose in `--voice`. Use a real
93	MiniMax `voice_id` (for example `male-qn-qingse`) or omit the flag
94	and let the daemon's default voice apply.
95	- Music: under 30s = single section; 30–90s = intro + body; 90s+ =
96	full arc. Don't try to fit a 3-act song into 15 seconds.
97	- SFX: prefer one well-described layer over a paragraph of "make it
98	cool" — generators reward specific texture words.
99	- Save the file every turn. The audio viewer shows transport controls
100	the moment the file lands.

People who liked this also viewed

6 picks

Type