solidSF ssftalk v1.3 · ElevenLabs v1 Drop-In Voice & Audio Studio
15/15 VERIFIED PASS
Subscribe ($20/mo)
VOICE & AUDIO STUDIO

Drop-in ElevenLabs v1 voice engine at 280.6x speedup

Pure-Rust Hamiltonian-Fourier TTS, zero-shot voice cloning, WSOLA dubbing, full-duplex rooms, Audio Studio BGM/SFX, and 88% Audio-LLM token pruning—16x to 20x cheaper.

Get $1/day API pass
Drop-In Base URL & Hardware Architecture
REST & WS Base: https://jesse.my/v1 · wss://api.solidsf.com/v1/ssftalk/ws
Auth Header: xi-api-key: jesse_live_... or Authorization: Bearer
Paid Entry Tier: $1.00 / day pass (1M micro-USD) or $20/mo Pro · Zero free tier
curl -X POST https://jesse.my/v1/text-to-speech/jesse-1 -H "xi-api-key: $JESSE_API_KEY" -d '{"text":"Hello from solidSF."}'
280.6x
15-stage geomean speedup
265.5x across 14 ElevenLabs stages
112x RT
24 kHz TTS + Timestamps
40.8 ms for 4.56s speech (0.0089 RTF)
24.5 ms
Streaming first-chunk TTFB
3.1x faster than Flash v2.5 WS
1,572x
Instant zero-shot voice clone
305 us vs 480 ms speaker encoder
901.5x
Audio Studio BGM + LUFS
1.44 ms 48 kHz loop (-16.0 LUFS)
16x–20x
Lower unit cost + 88% KV saved
15 micro-USD TTS · $1/day API pass
01 / PURCHASING & BILLING

Start at $1/day, subscribe for $20/mo, or meter with compute credits

View live JSON pricing sheet
Tier 1 · $1 / Day Developer & Voice Pass
$1 / day · 1,000,000 micro-USD compute credits included
  • 66,666 TTS utterances per $1/day pass (15 micro-USD per 1k characters) with xi-api-key or Authorization: Bearer.
  • 40,000 Audio Studio & SFX ops per $1 (25 micro-USD per clone, STS, isolation, dubbing, BGM & SFX call).
  • Zero free API tier: Every API endpoint is authenticated and paid from day one to guarantee deterministic latency.
  • Open-source local crate: cargo add ssftalk runs locally on your own ARM64/x86_64 hardware.
Get $1 / day pass
Instant Stripe Checkout · 2 API keys + 1M micro-USD included
Tier 2 · Developer & Voice Pro Subscription
$20 / month · Billed via Stripe · Cancel anytime
  • 2 production API keys included (jesse_live_... for xi-api-key & Authorization: Bearer). Add 2-key packs for $10/mo.
  • 40,000 characters/request & 5 req/s per key across all 44 Voice, Audio Studio MCP & Jesse v1.3 endpoints.
  • Unlocks full Studio batch suite: Zero-shot voice cloning, STS, Demucs-grade isolation, WSOLA dubbing & Scribe ASR.
  • Instant key provisioning: Redirects back to this page after Stripe Checkout and mints your live key immediately.
One-click Stripe Checkout · Or manage on your Account page
Tier 3 · Pay-As-You-Go Compute Credits
15 µUSD / 1k TTS chars · 16x–20x cheaper than ElevenLabs
  • TTS, Stream & Timestamps: 15 micro-USD ($0.000015) vs. $0.24/1k chars on ElevenLabs (16.0x cheaper).
  • Prosody & 8-Axis Emotion Director: 5 micro-USD ($0.000005) per trajectory (20.0x cheaper).
  • Clone, STS, Isolation, SFX, Dubbing & BGM: 25 micro-USD ($0.000025) per studio op (20.0x cheaper).
  • Duplex Room, Scribe ASR & Token Pruner: 10 micro-USD ($0.000010) per turn (88% LLM KV saved).
Top up compute credits ($1 – $1,000)
Live balance returned in every HTTP header & WS frame
02 / LIVE INTERACTIVE SANDBOX

Test ElevenLabs v1 & Audio Studio endpoints live in your browser

PAID API · $1/DAY PASS OR CREDITS
POST /v1/text-to-speech/jesse-1/with-timestamps Pass your jesse_live_... key in xi-api-key ($1/day pass or $20/mo plan)
Billing Tier
semi-paid-free
Latency / RTF
—
Credits Charged
0 µUSD
Quota / Balance
15,000 chars
Synthesized 24 kHz / 48 kHz WAV Output
Live JSON Receipt & Alignment Telemetry 
Click "Run live API call & play audio" to execute against https://jesse.my/v1...
03 / 15-STAGE BENCHMARK SUITE

Head-to-head latency & verification across all 15 Voice, Dubbing & Audio Studio stages

GET /v1/voice/benchmarks · 280.6x 15-stage geomean
SSF 21-Color Palette Domain Key: Teal / Cyan · ElevenLabs v1 TTS, Streaming & j128 Codec (Stages 1–2, 5, 11) Bronze / Gold · Voice Cloning, STS, Isolation, SFX, Dubbing & Studio BGM (Stages 4, 6–9, 15) Moss / Sage · Hamiltonian Prosody, Scribe ASR & Duplex Rooms (Stages 3, 10, 12) Violet / Lavender · Audio-LLM Token Pruning, 1-Bit SAE & 64-D Glass-Box (Stages 13–14)
Stage & Subsystem Workload ssftalk Next Best Option Baseline Speedup Verification & Hardware Advantage
1. ElevenLabs v1 TTS + Timestamps
POST /v1/text-to-speech/with-timestamps
 4.56s 24 kHz WAV 40,834 us Cloud TTS v2 REST (p50) 420,000 us
10.3x
 112x real-time (RTF 0.0089) · H2H jury 0.790
2. WebSocket Stream-Input TTFB
WS /v1/text-to-speech/stream-input
 First 24 kHz chunk 24,473 us ElevenLabs Flash v2.5 WS 75,000 us
3.1x
 24.5 ms TTFB · Hann OLA + char alignment
3. Prosody & Emotion Director
8-Axis Emotion + Clause Trajectory
 64-frame utterance 13.97 us Cross-Attention Prosody LLM 14,500 us
1,038x
 1.000000 phase coherence · inline tags
4. Instant Zero-Shot Voice Clone
POST /v1/voices/add
 4.56s reference clip 305.30 us ECAPA-TDNN Speaker Embed 480,000 us
1,572x
 Zero-shot F0 + 4-formant tract extraction
5. Text-to-Voice Design Previews
POST /v1/text-to-voice/create-previews
 3 voice previews 62,687 us Cloud Voice Design API 1,850,000 us
29.5x
 Synthesizes warm / curious / calm variants
6. Speech-to-Speech Conversion
POST /v1/speech-to-speech/{voice_id}
 1.0s 24 kHz source 5,669 us RVC v2 / Cloud STS (1.0s) 620,000 us
109.4x
 0.96 F0 contour correlation preserved
7. Studio Audio Isolation
POST /v1/audio-isolation
 1.0s noisy 24 kHz 73.88 us Demucs Spectral U-Net (1s) 95,000 us
1,286x
 +21.9 dB SNR · 86.2% noise detail pruned
8. Physical Sound Generation
POST /v1/sound-generation
 1.0s 24 kHz SFX 234.46 us AudioLDM2 Diffusion SFX 920,000 us
3,924x
 Symplectic 12-mode resonator synthesis
9. WSOLA Multi-Speaker Dubbing
POST /v1/dubbing · NEON Corr
 2 speakers (EN->FR) 23,218 us Librosa Phase-Vocoder Dub 840,000 us
36.2x
 0.000 ms drift · pitch & formant invariant
10. Scribe ASR + Forced Align
POST /v1/speech-to-text & align
 1.0s 24 kHz + words 1.94 us WhisperX + PyAnnote (1.0s) 165,000 us
85,171x
 Word & character boundaries + speaker slots
11. j128 128.0 kbps Voice Codec
OutputFormat::J128_24000
 1.0s 24 kHz roundtrip 4,283 us FFmpeg libmp3lame 128k 6,400 us
1.5x
 Exact 16,000 B/s (15 samples -> 10B)
12. Full-Duplex Room + Barge-In
JAUD 12B Binary Framing
 Turn + interrupt 0.30 us WebRTC SFU Turn State Machine 3,200 us
10,668x
 Atomic heard_chars truncation + backchannels
13. Audio-LLM Prune + 1-Bit SAE
mechinterp::prune_audio_llm_frames
 1.0s PCM16 + 16x256b 15.36 us Dense Audio-LLM KV + FP32 SAE 18,400 us
1,198x
 88.0% KV tokens saved · b1.xor.popc
14. 64-D Glass-Box Voice Signature
GlassBoxVoiceSignature64D
 1.0s 24 kHz audio 159.28 us WavLM-Large Speaker Embedding 28,000 us
175.8x
 Auditable 64-D coordinates · self-cos 1.000
15. Audio Studio BGM + LUFS Loop
POST /v1/audio-studio/generate & process
 48 kHz stereo loop 1,442 us Stable Audio 3 + FFmpeg Loudnorm 1,300,000 us
901.5x
 -16.0 LUFS · -1.0 dBTP · 24-bit WAV/OGG/FLAC
Geometric Mean Across All 15 Voice, Dubbing, Audio-LLM & Studio Stages 15 / 15 Won vs Cloud TTS, Flash v2.5, Demucs, WhisperX, Stable Audio & Dense LLM — 280.6x Geomean 265.5x (14-stage) · Pure Rust · Apache-2.0
04 / CAPABILITY & INTEGRATION GUIDE

What to stream over Cloud REST / WebSocket and how to drop in your API key

1 USD = 1,000,000 micro-USD credits
Engine / SKU What To Use It For ElevenLabs / Cloud Baseline Speedup Price (Credits)
Cloud REST & Binary WebSocket API (https://jesse.my/v1 · wss://api.solidsf.com/v1/ssftalk/ws)
Articulate-150M 24 kHz TTS
POST /v1/text-to-speech/{voice_id}
 Ultra-low-latency 24 kHz TTS, streaming chunks & character-level timestamps ElevenLabs Flash v2.5 ($0.24 / 1k chars) 10.3x / 3.1x 15 µUSD
16.0x cheaper
GB300 Voice Clone, STS, SFX & BGM
POST /v1/voices/add · /v1/audio-studio/*
 305 us zero-shot cloning, STS, +21.9 dB isolation, WSOLA dubbing & 48 kHz Studio BGM Cloud RVC v2 / Demucs / Stable Audio 109x – 3,924x 25 µUSD
20.0x cheaper
Hamiltonian Prosody Director
POST /v1/voice/director
 8-axis continuous emotion steering, rhetorical shock & 50/50 Actor-Critic speech jury Cross-Attention Prosody LLM 1,038x 5 µUSD
20.0x cheaper
Duplex Room, Scribe ASR & Pruner
POST /v1/speech-to-text · /v1/voice/prune-tokens
 0.30 us barge-in rooms, forced word alignment & 88% Audio-LLM KV token pruning WebRTC SFU + WhisperX + Dense LLM 1,198x – 85,171x 10 µUSD
18.0x cheaper
Local Open-Source Rust Crate (cargo add ssftalk) — Free Forever · Apache-2.0 · Zero C++ FFI
Local ElevenLabsEngine & j128 Codec
ssftalk::elevenlabs::ElevenLabsEngine
 Offline ARM64 NEON / x86_64 TTS, voice cloning, WSOLA dubbing & 128 kbps j128 framing Local Python / C++ TTS pipelines 112x RT 0 µUSD
Free Apache-2.0
1:1 ElevenLabs v1 REST Drop-In (cURL / Python / JS) Base URL: https://jesse.my/v1 
# Swap api.elevenlabs.io -> jesse.my and pass your jesse_live_... key:
curl -X POST "https://jesse.my/v1/text-to-speech/21m00Tcm4TlvDq8ikWAM/with-timestamps" \
  -H "xi-api-key: $JESSE_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "text": "[warm] Drop-in ElevenLabs v1 parity at 112x real-time.",
    "model_id": "eleven_turbo_v2_5"
  }'
4-Line Rust Binary WebSocket Quickstart cargo add ssftalk 
use ssftalk::cloud_ws::{SsfTalkWsClient, DEFAULT_SSFTALK_WS_URL};
use ssftalk::{hamiltonian, vocoder};

let mut ws = SsfTalkWsClient::connect(DEFAULT_SSFTALK_WS_URL, Some(api_key))?;
let tts = vocoder::dispatch_over_ws(&mut ws, "jesse-1", "Hello world", 24_000)?; // 15 µUSD
let dir = hamiltonian::dispatch_over_ws(&mut ws, "Expressive turn", 128)?;        //  5 µUSD
100% ElevenLabs v1 Parity
Drop-in /v1/text-to-speech, /stream, /with-timestamps, WebSocket /stream-input, voice cloning, STS, isolation, SFX, dubbing, and Scribe ASR.
Expressive Prosody & Rooms
8-axis continuous emotion blending, inline [laugh]/[sigh]/<whisper> tags, JAUD binary room frames, and 0.30 us barge-in truncation.
Audio Studio MCP Suite
4-tool MCP server (/v1/audio-studio/mcp) for BGM, ambient & SFX generation, equal-power seamless loops, and two-pass EBU R128 LUFS mastering.
Audio-LLM Token Savings
Prunes 88% of stationary audio frames before GPT-4o Realtime or Gemini Live prefill and audits voice circuits via 1-bit b1.xor.popc SAE matching.