Ross ROSS = Recommend OSS · open-source software intelligence for agents

FluidInference/FluidAudio

Frontier CoreML audio models in your apps — text-to-speech, speech-to-text, voice activity detection, and speaker diarization. In Swift, powered by SOTA open source. observed · 2026-08-28

github.com/FluidInference/FluidAudio · homepage · Swift · Apache-2.0 (permissive) observed · 2026-08-28

Health v2 · maintenance only

81/100

  • Activity 99
  • Release rhythm 86
  • Longevity 31
How is this computed?

round(0.45*activity + 0.35*rhythm + 0.20*longevity); archived -> min(score, 10) — computed 2026-09-03. Adoption (stars, forks) is never an input.

  • gap_med: 3.5
  • age_days: 438
  • days_rel: 14
  • days_push: 11
  • n_releases_24m: 63

Full methodology

Adoption not part of the score

2699 stars · 389 forks observed · 2026-08-28

What it is AI-extracted, prompt v1, taxonomy v1, 2026-08-30, confidence not recorded

A Swift SDK providing fully local, low-latency audio AI on Apple devices, including speech-to-text (Parakeet), speaker diarization (Pyannote/Sortformer), voice activity detection (Silero), and text-to-speech (Kokoro) via CoreML models. Inference is offloaded to the Apple Neural Engine to minimize CPU/GPU usage, with integrations for Swift Package Manager, CocoaPods, React Native, and Rust/Tauri.

Use cases

  • transcribe audio recordings to text on-device
  • add live streaming captions while the user speaks
  • identify who spoke when in meeting recordings
  • detect speech segments and skip silence before ASR
  • convert text to speech locally with multiple voices
  • recognize known speakers from pre-loaded voice profiles
  • build an always-on voice assistant on macOS or iOS

When to choose

  • you need private, fully local audio AI on Apple devices without cloud APIs
  • you want low-latency ASR, diarization, VAD, or TTS running on the Apple Neural Engine
  • you're building a Swift/iOS/macOS app and want a few-lines-of-code integration
  • you need streaming transcription or real-time speaker labels

When to avoid

  • you target Android, Windows, or Linux — this SDK is Apple-platform only
  • you need server-side or GPU-cluster inference rather than on-device
  • you need languages outside the 25 supported European languages for transcription
  • you need more than 4 speakers in streaming diarization (Sortformer limit)

Facets

library · maturity active

speech-recognition tts audio-processing machine-learning sdk speech-processing machine-learning apple-ecosystem cross-platform coreml apple-neural-engine speaker-diarization voice-activity-detection parakeet pyannote silero-vad kokoro-tts on-device-ai swift-package-manager audio macos ios swift gpu

8 sources

Member repositories

RepositoryRoleHealth v2
FluidInference/FluidAudiomain81

For agents

markdown · JSON · MCP: product_card(name="FluidInference/FluidAudio")

Data as of 2026-08-30T08:39:29.467469+00:00 · Report a problem