Text-to-Speech
ONNX
English
tts
speech-synthesis
neural-tts
offline-tts
on-device
edge-ai
edge
offline
cpu
onnxruntime
lightweight
small-model
flow-matching
vocos
english
raspberry-pi
flockspeech
conversational
conversational-ai
expressive-tts
nonverbal
laughter
disfluency
voice-assistant
game-audio
| Goldfinch v0.1 (flockspeech/goldfinch) | |
| Copyright 2026 Flock Speech | |
| Licensed under the Apache License, Version 2.0 (see LICENSE). This covers the Goldfinch acoustic model weights | |
| (onnx/encoder.onnx, onnx/decoder-int8.onnx, onnx-fp32/encoder.onnx, onnx-fp32/decoder.onnx) and the model | |
| configuration (onnx/goldfinch.json, onnx-fp32/goldfinch.json). | |
| Third-party components: | |
| - Vocos vocoder (onnx/vocoder.onnx, onnx-fp32/vocoder.onnx): converted from charactr/vocos-mel-24khz, | |
| Copyright (c) 2023 Charactr Inc., MIT License (see onnx/LICENSE-vocos). | |
| - Training data: DisfluencySpeech (amaai-lab/DisfluencySpeech), Apache License 2.0. Its texts are taken from | |
| Switchboard corpus transcripts; the Switchboard corpus is distributed by the Linguistic Data Consortium. | |
| Not included, required at run time to turn text into phonemes: | |
| - espeak-ng (GPL-3.0) and phonemizer (GPL-3.0). The flockspeech pip package installs them as separate | |
| dependencies. | |