jashansinghTT/s2-pro-blackhole

fish-s2-pro on Tenstorrent p150 and p300 and p300x2

Derived from fishaudio/s2-pro by Fish Audio, unmodified. See the original model card for training and evaluation details.

Fish Audio S2 Pro, a 4.5B dual-autoregressive text-to-speech model (Qwen3-based semantic decoder, 400M codebook decoder, 44.1 kHz DAC codec), with a fish-speech-compatible /v1/tts API (streaming WAV, voice cloning via references) and an OpenAI-compatible /v1/audio/speech endpoint.

Runs on Tenstorrent p150 or p300 or p300x2 or p100 — see the serving profiles below.

Packaged and published with tt-model-manager 0.1.0 (manifest schema 5.1).

At a glance

Hardware	p150, p300, p300x2
License	fish-audio-research-license

Intended use

Non-commercial speech synthesis from text, including streaming and cloning a voice from a short reference clip.

Prerequisites

• The Tenstorrent CLI, tt: uv tool install tenstorrent (or use tt-model alone, see the Quickstart)
• Docker
• Tenstorrent hardware: p150 or p300 or p300x2

Quickstart

tt serve jashansinghTT/s2-pro-blackhole

tt serve downloads the Docker image and the fishaudio/s2-pro weights at 1de9996b6be38b745688de084d87a5633f714e4e (into your HF cache; they are not in the image), then starts the model's own HTTP server on port 20000 (or the next free port, if that one is busy). The first start compiles kernels for your device, which takes several minutes; the server is ready when it logs Application startup complete.

Without tt-cli: tt-model serve jashansinghTT/s2-pro-blackhole.

Try it

curl -s localhost:20000/v1/tts -H 'Content-Type: application/json' \
  -d '{"text":"Hello from Tenstorrent.","format":"wav"}' -o hello.wav
curl -s localhost:20000/v1/audio/speech -H 'Content-Type: application/json' \
  -d '{"model":"fishaudio/s2-pro","input":"Hello.","voice":"default","response_format":"mp3"}' -o hello.mp3

Serving profiles

Select a different serving profile using --profile.

profile	hardware	mesh	description
p150 (default)	p150	P150	Single Blackhole chip (TP1). Widest compatibility.
p100	p100	P150	One p100, same settings as p150.
p300	p300	P300	One p300 board, both chips (TP2).
p300x2	p300x2	P300x2	Two p300 boards / QuietBox 2, 4 chips (TP4). Validated on the author's box.

Capabilities

This package exposes the model's own HTTP server on port 20000, not an OpenAI-compatible chat API — its request and response shapes are the model's own. See the author's notes for the payload it expects.

Voice cloning: POST /v1/references/add (multipart id / audio / text), then pass "reference_id" in /v1/tts.

Expected performance

eval	this package	reference	prompts
Whisper ASR word error rate	0.0	—	test clips
Semantic decoder output PCC vs PyTorch	>= 0.998 (1, 2 and 4 chips)	1.0	—

profile	metric	value
p150	real-time factor, non-streaming	~1.8 (6.3 s clip in 11.1 s; 36 s clip in 67 s)
p150	time to first audio, streaming /v1/tts	~0.4 s
p150	real-time factor, streaming	~1.9

Measured on one p150 chip with seeded requests to this package's server.

Limitations

• Slower than real time: about 1.8 s of compute per second of audio on one chip.
• The audio codec runs on the host CPU, not on the Tenstorrent device.
• Speech quality was checked on English text only.
• Non-commercial use only (Fish Audio Research License).

Feedback

Questions or problems with this package: open a discussion at https://huggingface.co/jashansinghTT/s2-pro-blackhole/discussions — that is what reaches its author. A problem with the tt tooling itself: tt report issue (collects your environment and opens a prefilled issue against tenstorrent/tt-cli). Product feedback: support@tenstorrent.com.

Provenance

The exact sources the image was built from — code/ in this repo is byte-identical to the model code inside the image:

component	built from
tt-metal	810e9006bd03d6f5f120a41cfa4758dc7fa1cfe1 (dirty tree — the image includes uncommitted changes)
code/ digest	f29564967e9fcf61 (sha256, first 16 hex digits)
built	2026-09-30T16:30:39+00:00 by tt-model 0.1.0