A small, expressive English text-to-speech model that laughs, breathes and says "uh", offline on any CPU. 44 MB, no GPU, no cloud.
Goldfinch is an 11.4M-parameter text-to-speech model from Flock Speech. Where
its sibling Junco reads aloud, Goldfinch talks: it keeps fillers such as
"uh" and "you know", restarts like "I, I think", and non-speech sounds you place with tags such as [laugh]. It runs
faster than real time on an ordinary CPU, with no GPU, network or PyTorch dependency.
Goldfinch is the conversational voice of the Flock Speech family, for chat companions, game characters,
conversational AI and voice assistants that should sound casual and human, on-device and offline. Install it with
pip install flockspeech.
| Flock Speech voices | Style | Model card |
|---|---|---|
| Goldfinch | conversational voice that laughs and breathes on cue with [laugh] tags |
this page |
| Junco | clear, neutral reading voice | flockspeech/junco |
| Developed by | Flock Speech |
| Model type | Neural text-to-speech: compact acoustic model + Vocos vocoder |
| Language | English (US) |
| Voices | 1 (conversational style) |
| Sound tags | [laugh], [breath], [sigh], [lipsmack], [throat], [noise] |
| Output | 24 kHz mono audio |
| Parameters | 11.4M acoustic model + 13.5M vocoder |
| Download | 44 MB (ONNX; int8 decoder, fp16 encoder and vocoder). A 100 MB full-precision build is also included |
| Runtime | ONNX Runtime on CPU (Linux, macOS, Windows; x86-64 and ARM64) |
| License | Apache-2.0 (vocoder weights MIT) |
| Version | v0.1, November 2026 |
All samples were made with the files in this repository at default settings (8 steps, seed 0), with no editing.
| Use | Text | Audio |
|---|---|---|
| Laugh | Oh my gosh, [laugh] I can't believe you actually did that. | |
| Chat | Yeah, you know, I used to work there, but, um, not anymore. | |
| Planning | [breath] Okay, so here's the plan for tomorrow. We meet at nine, and, uh, we'll figure out the rest. | |
| Reaction | That's hilarious. [laugh] No, really, it is. | |
| Plain | The weather today is sunny, with a light breeze from the west. |
The flockspeech Python package (0.2.0 or later) bundles these weights, so
nothing is downloaded at run time.
pip install flockspeech
flockspeech "Oh my gosh, [laugh] I can't believe you actually did that." --voice goldfinch -o laugh.wav --play
from flockspeech import Goldfinch
tts = Goldfinch() # loads in under a second
audio = tts("Yeah, um, we usually eat dinner around six. [laugh] Late, I know.") # float32, 24 kHz mono
tts.save("Well, you know, it depends.", "reply.wav")
for chunk in tts.stream(long_text): # sentence by sentence, for low latency
play(chunk)
To use the files in this repository directly:
from huggingface_hub import snapshot_download
from flockspeech import Goldfinch
path = snapshot_download("flockspeech/goldfinch", allow_patterns=["onnx/*"])
tts = Goldfinch(model_dir=f"{path}/onnx")
# or the full-precision build (100 MB, about 40% slower on a CPU)
path = snapshot_download("flockspeech/goldfinch", allow_patterns=["onnx-fp32/*"])
tts = Goldfinch(model_dir=f"{path}/onnx-fp32")
Write a tag in square brackets where the sound should go. Each tag is one symbol in the model's input, so the model decides how the sound is shaped and how long it lasts, in the context of the words around it.
| Tag | Also accepted | Training examples | Reliability |
|---|---|---|---|
[laugh] |
[laughs], [giggle], [chuckle] |
701 | good |
[breath] |
[inhale], [exhale] |
98 | good |
[throat] |
[clears throat], [cough] |
57 | fair |
[lipsmack] |
[lip smack] |
56 | fair |
[noise] |
19 | weak | |
[sigh] |
16 | weak |
By default the engine gives each tag 1.4× its predicted length and doubles the short pauses on either side. In
our listening comparison this made laughs sound smoother than the raw prediction. These settings are tag_len and tag_pad
in onnx/goldfinch.json.
Fillers need no tags: write them as words ("uh", "um", "you know", "I mean").
| Parameter | Default | Effect |
|---|---|---|
speed |
1.0 | Speaking rate; 1.2 is 20% faster |
steps |
8 | Generation steps; fewer is faster |
seed |
0 | Fixes the random variation in delivery, so output is reproducible |
threads |
up to 4 | CPU threads for ONNX Runtime |
Suitable for:
Out of scope:
[sigh] and [noise] had fewer than 20 training examples each, and can come out faint or wrong.seed if one take sounds off.Synthetic speech that laughs, breathes and hesitates sounds more human, which makes it easier to mistake for a real person. Goldfinch produces a single fixed voice and cannot clone speakers. We ask that you:
Text is converted to phonemes with espeak-ng, a compact neural acoustic model turns the phonemes into a mel spectrogram, and the Vocos vocoder (MIT) turns that into 24 kHz audio. All three run in ONNX Runtime on the CPU.
| File | Contents | Size |
|---|---|---|
onnx/encoder.onnx |
Acoustic model, part 1 (fp16 weights) | 7.2 MB |
onnx/decoder-int8.onnx |
Acoustic model, part 2 (int8) | 9.9 MB |
onnx/vocoder.onnx |
Vocos vocoder (fp16 weights) | 27.1 MB |
onnx/goldfinch.json |
Mel statistics, tag ids and default inference settings | <1 kB |
onnx-fp32/encoder.onnx |
Acoustic model, part 1, fp32 | 14.2 MB |
onnx-fp32/decoder.onnx |
Acoustic model, part 2, fp32 | 31.5 MB |
onnx-fp32/vocoder.onnx |
Vocos, fp32 | 54.2 MB |
onnx-fp32/goldfinch.json |
As above, pointing at the fp32 files | <1 kB |
samples/ |
The audio samples above |
DisfluencySpeech (AMAAI Lab, Apache-2.0): one studio speaker re-performing conversations from the Switchboard corpus transcripts, with fillers, restarts and non-speech events kept.
We trained on 8.3 hours of it.
| Compute | about 3.1 hours on one NVIDIA T4 GPU, from scratch |
Environmental impact. About 3.1 GPU-hours on one NVIDIA T4 (70 W TDP), or about 0.2 kWh.
How do I make it laugh?
Write [laugh] where the laugh should go: Goldfinch()("That's hilarious. [laugh] No, really."). The same works for
[breath], [throat], [lipsmack], [sigh] and [noise]; see Sound tags.
Do I need to tag "uh" and "um"? No. Write fillers as ordinary words ("uh", "um", "you know", "I mean") and Goldfinch says them naturally.
Does it need a GPU or an internet connection? No. It runs on an ordinary CPU with ONNX Runtime. Once installed, speech is generated fully offline and your text never leaves the machine.
How big is it? 44.2 MB in total: a 17.1 MB acoustic model plus the 27.1 MB Vocos vocoder, shared with Junco. Both voices together take 61.3 MB. A full-precision (fp32) build is also in this repository.
Can it run on a Raspberry Pi or other ARM boards?
It runs on 64-bit ARM Linux wherever onnxruntime and espeakng-loader publish wheels, including 64-bit Raspberry
Pi OS.
Does it support voice cloning or emotions? No voice cloning: Goldfinch is one English (US) voice. Expressiveness comes from fillers and the sound tags above, not from emotion labels.
Can I use it commercially? The weights are Apache-2.0 and the vocoder is MIT. The phonemiser (espeak-ng, phonemizer) is GPL-3.0 and is installed separately, and the training texts come from Switchboard transcripts; see the License section below.
onnx/vocoder.onnx): MIT, © 2023 Charactr Inc. See onnx/LICENSE-vocos.Run-time dependency. Text-to-phoneme conversion uses espeak-ng and phonemizer. Both are GPL-3.0 and are not included in this repository. See NOTICE.
@misc{flockspeech2026goldfinch,
title = {Goldfinch: a small conversational text-to-speech model with sound tags},
author = {{Flock Speech}},
year = {2026},
howpublished = {\url{https://huggingface.co/flockspeech/goldfinch}},
note = {Version 0.1}
}
Goldfinch is trained on DisfluencySpeech from the AMAAI Lab. We thank its authors and the authors of the Vocos vocoder.
onnx-fp32/vocoder.onnx 54 MBonnx-fp32/decoder.onnx 31 MBonnx/vocoder.onnx 27 MBonnx-fp32/encoder.onnx 14 MBonnx/decoder-int8.onnx 10 MBonnx/encoder.onnx 7 MB