/

/

clone higgs

clone higgs

AD

Over dit model

HIGGS AUDIO · Workflows

# Higgs TTS 3 (`audio-h34b`) - CLONE ONLY MODE **Higgs TTS 3** is built for voice chat—it speaks, not just reads. It turns model responses into expressive conversational speech across 100+ languages, featuring zero-shot voice cloning, character presets, and line-by-line control over emotion, style, prosody, pauses, and sound effects. Staff notes: quality and emotional range of your sample tends to lock you. If you have a sample of someone shouting and you try to use tags to make them whisper, it will struggle. It can return jobs several minutes in length. I got trump to give an entire lasagne recipe. BUT the voice segments if the render is too long, the voice output breaks in 15 to 20 second segments, depending on how you prompt. If you type long continuous sentences, higgs tends to segment itself regularly. If you provide strict grammar and tags to guide the voice, it tends to break less often. Clone-higgs also tends to inherit the ambient/room tone of your recording. If the person was outside in the wind when you took the recording, that carries over. --- ## 🛠 Syntax & Formatting Rules > ⚠️ **Important:** `` and `` formatted tags are deprecated. All control parameters must be defined at the **inpoint** of the prompt line using standard key-value syntax. ### Basic Syntax Structure /wf /run:audio-h34b tag1:value,tag2:value:: Your spoken dialogue text goes here. * **Tag Block Format:** `tag:value,tag:value::` * **Prompt Separator (`::`):** Marks the end of tag declarations and