moss-tts-v1.5-glade by Josh Leverette: Open-Weight Model

Model Card

By Josh Leverette, published under apache-2.0, revision 50f1447d2f48.

Hybrid conversion of MOSS-TTS-v1.5 and MOSS Audio Tokenizer. The 8.49B autoregressive language model runs on GPU through native Swift MLX; seventeen Core AI codec groups render mono 24 kHz PCM on ANE. English controls are qualified; other upstream languages are not qualified here. Glade is a native Swift library for speech inference on Apple silicon, using Core AI, Core ML, and model-specific host implementations. Its runtime source is currently private. This repository distributes the model assets and runtime metadata used by that library. Python is export/measurement tooling, not an application dependency. Mac qualification is included; iPhone and Watch qualification is not claimed.…

Read Josh Leverette's full model card

MOSS-TTS-v1.5 for Glade

Hybrid conversion of MOSS-TTS-v1.5 and MOSS Audio Tokenizer. The 8.49B autoregressive language model runs on GPU through native Swift MLX; seventeen Core AI codec groups render mono 24 kHz PCM on ANE. English controls are qualified; other upstream languages are not qualified here.

Glade is a native Swift library for speech inference on Apple silicon, using Core AI, Core ML, and model-specific host implementations. Its runtime source is currently private. This repository distributes the model assets and runtime metadata used by that library. Python is export/measurement tooling, not an application dependency. Mac qualification is included; iPhone and Watch qualification is not claimed. Download weights and configuration from one snapshot.

Layout and downloads

language/: mapped quantized language weights, tensor layout, configuration and tokenizer. Four-bit affine matrix groups, eight-bit audio heads and FP16 audio embeddings preserve the selected quantized runtime. No further training.
codec/: seventeen FP16 .aimodel groups plus original FP32 codebooks and boundary projections in boundary.f32/boundary.json.
voice-encoder/: optional original FP32 reference-audio encoder for importing client-supplied voices. Encoded conditioning can be cached and reused. Download this only when reference voice cloning is needed. No voice recording is included.

hf download coder543/moss-tts-v1.5-glade \
  --include 'language/*' 'codec/*' LICENSE README.md --local-dir ./moss

Glade's optional GladeMOSS product consumes language/ and codec/ directories. It supplies text cleanup, prompt construction, dynamic continuous batching, ordered incremental output, optional phase switching and MFA-assisted runout recovery. The planner targets five minutes per text unit with a hard limit of 5,000 generated positions; prompt positions are additional. Completed lanes leave the actual GPU batch. Memory budgets can reduce admitted lanes.

Mapped weights avoid repeated loading copies; clean pages are reclaimable, but mapping does not make the approximately 4.69 GiB of language weights disappear. The app must keep a mapped file unchanged while loaded. Generation retains each unit's full language history; codec blocks preserve their causal acoustic caches. Reference voice import can load the encoder before the language model and retain only encoded conditioning afterward. Neither path requires Python.

Measured Mac controls

M3 MacBook Air, 16 GB, macOS 27.0.1; Release runtime. These are bounded English functional controls, not a full Moon-speech benchmark or general perceptual study.

A longer GPU-language/ANE-codec control generated 48.40 s of audio in 37.91 s (1.28×), excluding preparation and WAV writing. Its language frames exactly matched the separate GPU-codec control.
Two complete 197-word units produced 157.36 s of audio. Source-ordered online phased and resident rendering produced identical PCM and language tokens.
A deliberately forced runout retained verified audio through a sentence at 25.63 s after 20.16 s was delivered, then generated only the unspoken suffix. The other batched sequence remained intact. Cohere recovered all 394 normalized words from both this output and the ordinary control.
The original FP32 codec comparison on a 78.48-second token stream differs by 5.60% relative waveform RMS. Streaming/phase policies match each other exactly; the converted codec is not numerically lossless against FP32.

Initial observed preparation of the seventeen codec assets took 30.51 s; subsequent cached loading took 0.24 s. Related compiler caches may have been reused. These values exclude language loading/prefill, and source-location stripping changes artifact identity. They are not guaranteed fresh-device startup times. Mapped load time can also defer work to the first GPU call. Memory accounting must include mapped clean pages, GPU allocations, codec allocations and KV caches. No phone memory allowance, general MOS or power-efficiency claim is made.

Attribution

Language source revision: cdd3b911b1585e3f2dbc7775ef10f9926f58850a. Codec/voice source revision: 3cd226ba2947efa357ef453bcad111b6eafba782. Original weights/source use Apache-2.0; see LICENSE and the two upstream cards. Authoring source locations are stripped from Core AI assets while graph signatures and operation statistics are checked unchanged. No compiled specialization cache is distributed.

Identity and Version

Repository: coder543/moss-tts-v1.5-glade
Publisher: Josh Leverette
Task: Text to speech
Modality: Audio
Library: mlx
Parameters: Not stated by the source
Languages: en
Revision: 50f1447d2f48c57738306a68c872ec9055e5ee0b
First published: 2026-10-09
Last updated: 2026-10-09

Files and Weights

132 files, 10.4 GB in total. The weights are 70 files totalling 8.6 GB in bin, safetensors.

Weights70 files · 8.6 GB

Configuration22 files · 141.2 KB

Tokenizer1 file · 11.4 MB

Documentation3 files · 16.4 KB

Other35 files · 1.8 GB

Repository1 file · 2.8 KB

Every file

File	Type	Size	SHA-256
language/weights.bin	Weights	5.0 GB	828c06b2ccdd
voice-encoder/encoder-1-0.safetensors	Weights	28.3 MB	b99c16b0f2ea
voice-encoder/encoder-1-1.safetensors	Weights	28.3 MB	0275d160e0d2
voice-encoder/encoder-1-10.safetensors	Weights	28.3 MB	5f41bc10b082
voice-encoder/encoder-1-11.safetensors	Weights	28.3 MB	2092e40c4b6d
voice-encoder/encoder-1-2.safetensors	Weights	28.3 MB	56b8625206ef
voice-encoder/encoder-1-3.safetensors	Weights	28.3 MB	3f4d05f7d68b
voice-encoder/encoder-1-4.safetensors	Weights	28.3 MB	aafbc640ac6c
voice-encoder/encoder-1-5.safetensors	Weights	28.3 MB	83a6a4a37d18
voice-encoder/encoder-1-6.safetensors	Weights	28.3 MB	de435bebb467
voice-encoder/encoder-1-7.safetensors	Weights	28.3 MB	731493c5015a
voice-encoder/encoder-1-8.safetensors	Weights	28.3 MB	cebcbbf30f5e
voice-encoder/encoder-1-9.safetensors	Weights	28.3 MB	02ab4379436d
voice-encoder/encoder-3-0.safetensors	Weights	28.3 MB	30387e9193b6
voice-encoder/encoder-3-1.safetensors	Weights	28.3 MB	c97c3b3e75ab
voice-encoder/encoder-3-10.safetensors	Weights	28.3 MB	59870c714bc4
voice-encoder/encoder-3-11.safetensors	Weights	28.3 MB	c60fc36f6706
voice-encoder/encoder-3-2.safetensors	Weights	28.3 MB	dc2d86e03dc7
voice-encoder/encoder-3-3.safetensors	Weights	28.3 MB	e31b476ef74f
voice-encoder/encoder-3-4.safetensors	Weights	28.3 MB	7fd276a1dbaa
voice-encoder/encoder-3-5.safetensors	Weights	28.3 MB	c332081e4cad
voice-encoder/encoder-3-6.safetensors	Weights	28.3 MB	c9aa57a496a3
voice-encoder/encoder-3-7.safetensors	Weights	28.3 MB	e4b0a5c63d57
voice-encoder/encoder-3-8.safetensors	Weights	28.3 MB	2f3e77af3bd3
voice-encoder/encoder-3-9.safetensors	Weights	28.3 MB	516b9ef9cfbb
voice-encoder/encoder-5-0.safetensors	Weights	28.3 MB	730e01343e74
voice-encoder/encoder-5-1.safetensors	Weights	28.3 MB	6a21dd563b9f
voice-encoder/encoder-5-10.safetensors	Weights	28.3 MB	020f00e56687
voice-encoder/encoder-5-11.safetensors	Weights	28.3 MB	ec237a090a5a
voice-encoder/encoder-5-2.safetensors	Weights	28.3 MB	4f6d076993ad
voice-encoder/encoder-5-3.safetensors	Weights	28.3 MB	acd140a564d2
voice-encoder/encoder-5-4.safetensors	Weights	28.3 MB	6113e90d1a78
voice-encoder/encoder-5-5.safetensors	Weights	28.3 MB	9291f7c59746
voice-encoder/encoder-5-6.safetensors	Weights	28.3 MB	e8e07ffea53f
voice-encoder/encoder-5-7.safetensors	Weights	28.3 MB	7167c462100d
voice-encoder/encoder-5-8.safetensors	Weights	28.3 MB	09494678ff5e
voice-encoder/encoder-5-9.safetensors	Weights	28.3 MB	a8a9d08eb73c
voice-encoder/encoder-7-0.safetensors	Weights	78.7 MB	27141a355827
voice-encoder/encoder-7-1.safetensors	Weights	78.7 MB	0b48d0a662c5
voice-encoder/encoder-7-10.safetensors	Weights	78.7 MB	4010b9028393
voice-encoder/encoder-7-11.safetensors	Weights	78.7 MB	7c1804eda94f
voice-encoder/encoder-7-12.safetensors	Weights	78.7 MB	6c752b0ccb75
voice-encoder/encoder-7-13.safetensors	Weights	78.7 MB	8b9c915e0b62
voice-encoder/encoder-7-14.safetensors	Weights	78.7 MB	e0b64e33152d
voice-encoder/encoder-7-15.safetensors	Weights	78.7 MB	02a6fa06e4ed
voice-encoder/encoder-7-16.safetensors	Weights	78.7 MB	706b1c79a055
voice-encoder/encoder-7-17.safetensors	Weights	78.7 MB	08899a9e05b5
voice-encoder/encoder-7-18.safetensors	Weights	78.7 MB	9a1e5421a4ce
voice-encoder/encoder-7-19.safetensors	Weights	78.7 MB	1cb0363a3fd1
voice-encoder/encoder-7-2.safetensors	Weights	78.7 MB	446e1ee95003
voice-encoder/encoder-7-20.safetensors	Weights	78.7 MB	73b250a64d20
voice-encoder/encoder-7-21.safetensors	Weights	78.7 MB	f079fb67fe97
voice-encoder/encoder-7-22.safetensors	Weights	78.7 MB	9e6a8b869a59
voice-encoder/encoder-7-23.safetensors	Weights	78.7 MB	e1daa293a89e
voice-encoder/encoder-7-24.safetensors	Weights	78.7 MB	dbe87166d5d9
voice-encoder/encoder-7-25.safetensors	Weights	78.7 MB	887c2c9b5476
voice-encoder/encoder-7-26.safetensors	Weights	78.7 MB	df87095165bb
voice-encoder/encoder-7-27.safetensors	Weights	78.7 MB	b47b04a050dd
voice-encoder/encoder-7-28.safetensors	Weights	78.7 MB	908ea40a05d2
voice-encoder/encoder-7-29.safetensors	Weights	78.7 MB	7a19489627e6
voice-encoder/encoder-7-3.safetensors	Weights	78.7 MB	69a32e7ae9a9
voice-encoder/encoder-7-30.safetensors	Weights	78.7 MB	f4455c9c9287
voice-encoder/encoder-7-31.safetensors	Weights	78.7 MB	061986af7de3
voice-encoder/encoder-7-4.safetensors	Weights	78.7 MB	01339324815c
voice-encoder/encoder-7-5.safetensors	Weights	78.7 MB	95e3e989195f
voice-encoder/encoder-7-6.safetensors	Weights	78.7 MB	400f38dc86e3
voice-encoder/encoder-7-7.safetensors	Weights	78.7 MB	0bbab4d134d6
voice-encoder/encoder-7-8.safetensors	Weights	78.7 MB	5ae17434f4ed
voice-encoder/encoder-7-9.safetensors	Weights	78.7 MB	443fc83c4d8a
voice-encoder/projections.safetensors	Weights	9.0 MB	0291e0f0d706
codec/boundary.json	Configuration	21.6 KB	—
codec/codec-0-0.aimodel/metadata.json	Configuration	105 B	—
codec/codec-0-1.aimodel/metadata.json	Configuration	105 B	—
codec/codec-0-2.aimodel/metadata.json	Configuration	105 B	—
codec/codec-0-3.aimodel/metadata.json	Configuration	105 B	—
codec/codec-0-4.aimodel/metadata.json	Configuration	105 B	—
codec/codec-0-5.aimodel/metadata.json	Configuration	105 B	—
codec/codec-0-6.aimodel/metadata.json	Configuration	105 B	—
codec/codec-0-7.aimodel/metadata.json	Configuration	105 B	—
codec/codec-2-0.aimodel/metadata.json	Configuration	105 B	—
codec/codec-2-1.aimodel/metadata.json	Configuration	105 B	—
codec/codec-2-2.aimodel/metadata.json	Configuration	105 B	—
codec/codec-4-0.aimodel/metadata.json	Configuration	105 B	—
codec/codec-4-1.aimodel/metadata.json	Configuration	105 B	—
codec/codec-4-2.aimodel/metadata.json	Configuration	105 B	—
codec/codec-6-0.aimodel/metadata.json	Configuration	105 B	—
codec/codec-6-1.aimodel/metadata.json	Configuration	105 B	—
codec/codec-6-2.aimodel/metadata.json	Configuration	105 B	—
codec/metadata.json	Configuration	5.1 KB	—
language/config.json	Configuration	2.3 KB	—
language/mapped-weights.json	Configuration	106.1 KB	—
voice-encoder/metadata.json	Configuration	4.4 KB	—
LICENSE	Documentation	11.4 KB	—
NOTICE	Documentation	158 B	—
README.md	Documentation	4.9 KB	—
codec/boundary.f32	Other	14.3 MB	95784f39fa7c
codec/codec-0-0.aimodel/main.hash	Other	32 B	—
codec/codec-0-0.aimodel/main.mlirb	Other	157.4 MB	d35e3639bf4d
codec/codec-0-1.aimodel/main.hash	Other	32 B	—
codec/codec-0-1.aimodel/main.mlirb	Other	157.4 MB	0b68c1d71355
codec/codec-0-2.aimodel/main.hash	Other	32 B	—
codec/codec-0-2.aimodel/main.mlirb	Other	157.4 MB	6d02e094ef27
codec/codec-0-3.aimodel/main.hash	Other	32 B	—
codec/codec-0-3.aimodel/main.mlirb	Other	157.4 MB	eeb8630fd649
codec/codec-0-4.aimodel/main.hash	Other	32 B	—
codec/codec-0-4.aimodel/main.mlirb	Other	157.4 MB	5d636b1aeb9c
codec/codec-0-5.aimodel/main.hash	Other	32 B	—
codec/codec-0-5.aimodel/main.mlirb	Other	157.4 MB	ac61bc947261
codec/codec-0-6.aimodel/main.hash	Other	32 B	—
codec/codec-0-6.aimodel/main.mlirb	Other	157.4 MB	861c1ad03890
codec/codec-0-7.aimodel/main.hash	Other	32 B	—
codec/codec-0-7.aimodel/main.mlirb	Other	157.4 MB	f3fc20a1db4e
codec/codec-2-0.aimodel/main.hash	Other	32 B	—
codec/codec-2-0.aimodel/main.mlirb	Other	56.7 MB	6b76db2e85ab
codec/codec-2-1.aimodel/main.hash	Other	32 B	—
codec/codec-2-1.aimodel/main.mlirb	Other	56.7 MB	10c7ddba5a2b
codec/codec-2-2.aimodel/main.hash	Other	32 B	—
codec/codec-2-2.aimodel/main.mlirb	Other	56.7 MB	879d022d5999
codec/codec-4-0.aimodel/main.hash	Other	32 B	—
codec/codec-4-0.aimodel/main.mlirb	Other	56.7 MB	86b040459bd4
codec/codec-4-1.aimodel/main.hash	Other	32 B	—
codec/codec-4-1.aimodel/main.mlirb	Other	56.7 MB	22a52ffe485a
codec/codec-4-2.aimodel/main.hash	Other	32 B	—
codec/codec-4-2.aimodel/main.mlirb	Other	56.7 MB	d0be194b0f26
codec/codec-6-0.aimodel/main.hash	Other	32 B	—
codec/codec-6-0.aimodel/main.mlirb	Other	56.7 MB	2b62a7abfbfe
codec/codec-6-1.aimodel/main.hash	Other	32 B	—
codec/codec-6-1.aimodel/main.mlirb	Other	56.7 MB	51130f58ded0
codec/codec-6-2.aimodel/main.hash	Other	32 B	—
codec/codec-6-2.aimodel/main.mlirb	Other	56.7 MB	222f98183ec3
.gitattributes	Repository	2.8 KB	—
language/tokenizer.json	Tokenizer	11.4 MB	cb3c8fa82993

License and Download

License: apache-2.0
Access: Open weights, no gate
Download size: 8.6 GB

Download from Josh Leverette

Released by Josh Leverette through its official repository on Hugging Face. Read the license.

Built From

Derived from OpenMOSS-Team/MOSS-Audio-Tokenizer
Quantized from OpenMOSS-Team/MOSS-Audio-Tokenizer

Memory Requirements

Precision	Weights in memory
As published	8.6 GB

Weights only, from the published parameter count; the key-value cache and runtime add to this.

Questions About moss-tts-v1.5-glade

Can I use moss-tts-v1.5-glade commercially?

Yes. moss-tts-v1.5-glade is released under Apache License 2.0. The Apache License 2.0 is a permissive open-source license. It permits commercial use, modification and redistribution. It requires keeping the license and copyright notices and any NOTICE file, stating significant changes, and it includes an express patent grant from contributors.

Similar Models

Model · Text to speech

Kokoro-82M

Hexgrad

Kokoro is an open-weight TTS model with 82 million parameters. Despite its lightweight architecture, it delivers comparable quality to larger models while being significantly faster and more cost-efficient. With Apache-licensed weights, Kokoro can be deployed anywhere from production environments to personal projects. You can run this basic cell on Google Colab. Listen to samples. For more languages and details, see Advanced Usage. Under the hood, kokoro uses misaki, a G2P library at https://github.com/hexgrad/misaki Model SHA256 Hash: 496dba118d1a58f5f3db2efc88dbdc216e0483fc89fe6e47ee1f2c53f18ad1e4 Data: Kokoro was trained exclusively on permissive/non-copyrighted audio data and IPA…

Open weights apache-2.0

View model

Model · Text to speech

audio.cpp-gguf

Audio.cpp

This directory contains audio.cpp-native GGUF conversions of multiple speech models. These files are intended for use with audio.cpp. If you enjoy the project, please star audio.cpp on GitHub and this Hugging Face repository. For conversion details, supported layouts, direct-file loading, sidecar embedding, and the latest compatibility notes, see the audio.cpp GGUF guide: - https://github.com/0xShug0/audio.cpp/blob/main/docs/gguf.md!!! Converted and quantized packages are checked with automated metrics, but perceived quality can still differ for human listeners. Please validate the exact package, backend, and route to confirm the output is acceptable for your use case. The table lists the…

Open weights other audio.cpp

View model

Model · Text to speech

XTTS-v2

Coqui.ai

ⓍTTS is a Voice generation model that lets you clone voices into different languages by using just a quick 6-second audio clip. There is no need for an excessive amount of training data that spans countless hours. This is the same or similar model to what powers Coqui Studio and Coqui API. - Supports 17 languages. - Voice cloning with just a 6-second audio clip. - Emotion and style transfer by cloning. - Cross-language voice cloning. - Multi-lingual speech generation. - 24khz sampling rate. - 2 new languages; Hungarian and Korean - Architectural improvements for speaker conditioning. - Enables the use of multiple speaker references and interpolation between speakers. - Stability…

Open weights other coqui

View model

Model · Text to speech

chatterbox

Resemble AI

Chatterbox Multilingual V3 is the latest general-purpose multilingual TTS model in the Chatterbox family. It keeps the same 0.5B model size while improving speaker similarity, reducing hallucinations, and producing more natural, conversational speech across languages. V3 is designed for broad language coverage like V2, but with stronger stability and more expressive generation. It is the recommended multilingual model for users who want one voice cloning model that works across many languages. Try it in the Chatterbox Multilingual TTS V3 Space. Alongside V3, we are releasing the Single Language Pack: dedicated finetunes for priority languages where tighter quality control, stronger…

Open weights mit chatterbox

View model

Model · Text to speech

F5-TTS

Yushen CHEN

Download F5-TTS or E2 TTS and place under ckpts/ Paper: F5-TTS: A Fairytaler that Fakes Fluent and Faithful Speech with Flow Matching

Open weights cc-by-nc-4.0 f5-tts

View model

Model · Text to speech

Kokoro-82M-v1.0-ONNX

ONNX Community

Kokoro is a frontier TTS model for its size of 82 million parameters (text in/audio out). First, install the kokoro-js library from NPM using: You can then generate speech as follows: Optionally, save the audio to a file: The model is resilient to quantization, enabling efficient high-quality speech synthesis at a fraction of the original model size.

Open weights apache-2.0 transformers.js

View model