This is a collection of workflows for the Qwen3TTS speech large model, consisting of a total of 6 functions. Additionally, there is a translation node (not applicable locally). You can experience it online ：https://www.runninghub.ai/post/2015774303727390721/?inviteCode=rh-v1401
or deploy it locally.
Functions: Voice cloning, sound design, sound customization, emotional enhancement, multi-character
About Voice Design ,how to make you own voice:
Description of sound:
1.Basic acoustic
features Gender/Age: Youth, Middle-aged, Mature, Elderly, Adolescent, Childlike voice. Pitch: High-pitched, Low-pitched, Sweet, Rich, Clear. Timbre: Bright, Warm, Rough, Magnetic, Clear, Grainy. Speech rate and rhythm: Light, Steady, Rapid, Smooth, Rhythmic.
2. Language Expression Characteristics
The way of speaking, tone and emotions. Emotions and tones: cheerful, gentle, calm, serious, passionate, lazy, melancholic, encouraging, excited. Pronunciation features: accurate and melodious, clear, with a slight accent (such as a southern accent), languid, crisp and neat. Expression styles: friendly, professional, intellectual, lively, composed, persuasive, storytelling-like. 3. Scenes and Characters
The context in which the voice is used helps the model determine the appropriate role. Occupation/Role: News anchor, teacher, doctor, commentator, friend, guide, intelligent assistant, movie narrator. Usage Scenarios: Children's stories, commercial advertisements, classroom lectures, game characters.
About emotion :
happy and naturally are way more useful,
More workflow in My Runinghub homepage , free to try and download ,all I need is your rencouragement and recognition !!!
Description
Qwen3TTS 1.7B model ，3 models include
FAQ
Comments (4)
what about audio to audio? using an audio file to give the emotion, cadence tempo, etc to the voice of a pth model
That I would like to use the Index TTX to do it , next workflow I can show you how to do~~
@Adventurer17 Can this replace the original singer's voice with my voice?
@Rixh666 The current open-source TTS models all perform poorly in singing. Recently, there is an open-source voice substitution model called soulX, but the effect is mediocre. I haven't tested it yet. If I find it good after testing, I will upload it to Civitai.