Vantage Step Audio EditX
This project is a custom node implementation built on top of Step-Audio-EditX. It adapts and extends EditX capabilities to support multi‑speaker, long‑format, voice cloning, and emotion/style/speed editing, enabling you to feed in a script with multiple speakers, inline pauses, paralinguistic cues, and get a concatenated audio output in one pass.
Install / Use
npx skills add vantagewithai/Vantage-Step-Audio-EditXInstalls into whichever agent you are using.
README
Step‑Audio‑EditX Multi‑Voice Cloner Node 🎙️
This project is a custom node implementation built on top of Step-Audio-EditX. It adapts and extends EditX capabilities to support multi‑speaker, long‑format, voice cloning, and emotion/style/speed editing, enabling you to feed in a script with multiple speakers, inline pauses, paralinguistic cues (like laughter, breathing), and get a concatenated audio output — all in one pass.
Description
The original Step‑Audio‑EditX model enables single‑speaker voice cloning and emotion/style editing given a reference prompt audio + text.
This node extends that capability, allowing you to:
- Provide multiple “speaker” reference voices at once.
- Write a simple script with speaker tags, inline pauses, and optionally emotion/style/speed tags.
- Generate a single contiguous audio file with all voices, pauses, and editing applied.
- Handle paralinguistic markers (like
[Laughter],[Breathing], etc.) — these are preserved and synthesis attempts to reflect them as natural speech or silence, depending on your engine’s capabilities.
In short: you can build multi‑voice dialogues, audio stories, podcasts, or voice‑over sequences in one go.
Features
- Multi‑speaker support (map each speaker to a reference audio + prompt).
- Inline speaker switching via
[speakerX]tags. - Inline pauses via
[pause]N]syntax (pause of N milliseconds). - Emotion / style / speed tags (e.g.
[happy],[serious],[faster]) for each line. - Paralinguistic tag support — e.g.
[Laughter],[Breathing],[Sigh],[Dissatisfaction-hnn], etc. Those tags remain in the output text. - Automatic concatenation of generated audio segments into one final waveform.
- Progress reporting (with progress bar).
- Graceful handling of missing speaker‑tags (defaults to first speaker).
How It Works
- Parse the input script line by line.
- Detect tags:
[speakerX]— which reference voice/prompt to use.- Optional leading tags like emotion, style, speed (e.g.
[happy],[whisper],[slower]). - Paralinguistic tags (preserved).
[pause]tags — interpreted as “generate silence for N ms.”
- For each speech line, call
clone_from_tensor(...)(and optionally repeated editing for emotion / style / speed). - For pause lines, generate a tensor of zeros of the requested duration.
- Collect all segments (speech or silence), concatenate them, and return a single audio output.
Usage
Example usage in a ComfyUI flow:
[speaker1][happy]Hello there!
[pause]500
[speaker2][sad][whisper]I’m not sure about this…
[speaker1][Laughter]That’s hilarious!
-
Provide reference audios & prompts for each speaker.
-
Feed this script to the node.
-
Get a single AUDIO output: concatenated waveform with cloned voices, pauses, and editing.
-
Emotion and Speaking Style Editing
- Remarkably effective iterative control over emotions and styles, supporting dozens of options for editing.
- Emotion Editing : [ Angry, Happy, Sad, Excited, Fearful, Surprised, Disgusted, etc. ]
- Speaking Style Editing: [ Act_coy, Older, Child, Whisper, Serious, Generous, Exaggerated, etc.]
- Editing with more emotion and more speaking styles is on the way. Get Ready! 🚀
- Remarkably effective iterative control over emotions and styles, supporting dozens of options for editing.
-
Paralinguistic Editing
- Precise control over 10 types of paralinguistic features for more natural, human-like, and expressive synthetic audio.
- Supporting Tags:
- [ Breathing, Laughter, Suprise-oh, Confirmation-en, Uhm, Suprise-ah, Suprise-wa, Sigh, Question-ei, Dissatisfaction-hnn ]
-
Available Tags
Related Skills
gortex
1.1kHigh-performance code-intelligence engine for AI agents and IDE, supports 257 languages, multi repositories, based on graph, with access via CLI, MCP Server, and API. AI coding agents teammate - expose only needed information, cutting token usage up to 50x. 100% local.
techrogue
TechRogue – Roguelike technical quiz for engineers. Usage: /techrogue | /techrogue build | /techrogue settings
cc-switch
125.6kA cross-platform desktop All-in-One assistant for Claude Code, Codex, OpenCode, OpenClaw, Grok Build & Hermes Agent. Only official website: ccswitch.io
cc-switch
125.7kA cross-platform desktop All-in-One assistant for Claude Code, Codex, OpenCode, OpenClaw, Grok Build & Hermes Agent. Only official website: ccswitch.io
