Directable, multi-speaker speech
The Text to Speech node runs S2.1 Pro with up to five voices. Tag speakers as @Voice1 and @Voice2, then direct each line with cues like [whispering] or [excited].
Fish Audio voices in ComfyUI
Fish Audio ships as official Partner Nodes in ComfyUI. Generate expressive speech with S2.1 Pro, clone a voice from a short recording and transcribe audio, all in the same graph as your image and video models.
CategoryCreative workflows
IntegrationOfficial Partner Nodes in ComfyUI
Where to find itNode menu → partner/audio/Fish Audio
Default modelS2.1 Pro, up to five voices per node
NodesText to Speech, Speech to Text, Voice Selector, Instant Voice Clone
TemplatesText to Speech, Voice Clone to Speech, Speech to Text, on Comfy Cloud or in the Template Library
Voice cloningFree in the Instant Voice Clone node; you pay only for generated speech
Fish Audio Partner Nodes in ComfyUI ↗Why Comfy and Fish Audio
ComfyUI is where creators wire image, video and 3D models into repeatable workflows. Fish Audio adds the voice: expressive, directable speech and cloning that drops into the same graph, so a scene can go from script to finished audio without leaving Comfy.
Clone a voice inside the graph
The Instant Voice Clone node turns a short recording into a private voice you can reuse across workflows. Cloning is free; you only pay for the speech you generate.
Localize a scene from one workflow
S2.1 Pro speaks 30+ languages, so one graph can voice a scene for every market. Swap the script, keep the voice, and render each version in one run.
How it works
3 steps from a Fish Audio voice to a shipping product on Comfy.
STEP 01
Add the Fish Audio nodes
Update to the latest ComfyUI, then open a Fish Audio template (Text to Speech, Voice Clone to Speech or Speech to Text) or add the nodes from partner/audio/Fish Audio.
STEP 02
Sign in and pick a voice
Sign in to your ComfyUI account when prompted, then connect a preset voice, any fish.audio voice model, or a fresh clone from Instant Voice Clone.
STEP 03
Write, run and save
Write the script with cues like [whispering] or [laughing], run the workflow, then save the audio or pass it straight to your video and lip-sync nodes.
What teams build with Fish Audio and Comfy
Talking characters and avatars
Pair Fish Audio speech with image-to-video and lip-sync nodes to give every character a consistent voice, in one graph.
Multi-character dialogue
Cast up to five voices in a single Text to Speech node for scenes, shorts, trailers and podcasts.
Transcripts and captions
The Speech to Text node returns text, the detected language and timed segments, ready for subtitles or the next step in the workflow.
More partners
AI video platform. Fish Audio powers the Fish voice engine inside HeyGen's AI Studio, with accent and locale controls for English. Available in both HeyGen Video Creator and LiveAvatar.
Video & avatars
AI Gateway for every model behind one key. Fish Audio S2.1 Pro, S2 Pro, S1 and Transcribe-1 are all available as fish-audio/ models.
Routers & gateways
Unified API for hundreds of models. Fish Audio S2.1 Pro is available as fish-audio/s2.1-pro through OpenRouter's text-to-speech API.
Routers & gateways
Start building with Fish Audio and Comfy
See how the Fish Audio Partner Nodes work in ComfyUI, or talk to our partnerships team about creator programs and co-marketing.