Releases · OpenBMB/VoxCPM

Release list

Latest

a710128 released this 11 May 11:59

19b6bf7

This release focuses on fine-tuning usability, runtime stability, safer LoRA loading, and faster streaming inference.

Added voxcpm validate for pre-flight JSONL training manifest validation.
Added optional ref_audio support in the fine-tuning data pipeline.
Improved runtime device handling with explicit --device support and safer MPS dtype behavior.
Improved VoxCPM2 streaming VAE decoding by avoiding redundant overlap decoding.
Hardened legacy LoRA checkpoint loading with weights_only=True.
Fixed LoRA rank mismatch handling in lora_ft_webui.py.

Add voxcpm validate --manifest train.jsonl to catch training data issues before fine-tuning.
- Validates JSONL format, required text/audio fields, audio existence/readability, sample rate, duration stats, text length stats, and optional ref_audio.
Add optional ref_audio support for fine-tuning manifests.
- Training packing now supports [103, ref_audio, 104, text, 101, target_audio, 102].
- Loss is applied only to the target audio segment.
Add --device CLI argument for model inference commands.
- Supports auto, cpu, mps, cuda, and indexed CUDA devices such as cuda:0.

Improve VoxCPM2 streaming VAE decode with a stateful StreamingVAEDecoder.
- Streaming decode now processes only the newest latent patch and carries causal convolution state internally.
- This removes redundant overlap decoding and reduces streaming VAE decode overhead.

Fix CUDA Graph dynamic-shape accumulation by using the uncompiled feature encoder for prefill.
Fix CPU SDPA attention mask broadcasting by using an explicit broadcastable mask shape.
Fix non-string text validation order to raise the intended ValueError instead of AttributeError.
Fix file descriptor leaks when loading config.json in local model loaders.
Fix MPS audio quality issues by promoting low-precision dtypes to float32 on Apple Silicon by default.
Fix VOXCPM_MPS_DTYPE override validation to match supported dtype aliases.
Fix LoRA rank mismatch in lora_ft_webui.py by reloading the model when checkpoint rank differs.
Fix Web Demo control text handling by stripping parentheses before constructing the model prompt.

Legacy LoRA .ckpt / .pth loading now uses torch.load(..., weights_only=True).
This reduces the risk of arbitrary pickle payload execution while preserving tensor-only checkpoint compatibility.

Added coverage for training manifest validation, including sample-rate mismatch, missing audio, relative paths, ref_audio, and CLI exit codes.
Added runtime device selection tests.
Added LoRA checkpoint safety tests for tensor-only checkpoints and malicious pickle payloads.
Added CLI tests for --device defaults and argument forwarding.

Thanks to the contributors included in this release:

Full Changelog: 2.0.2...2.0.3

Oumnya, SuperMarioYL, and 6 other contributors

Assets 2

v2.0.2: fixed some bugs

a710128 released this 08 Apr 10:19

68af4fe

Full Changelog: 2.0.1...2.0.2

Assets 2

a710128 released this 08 Apr 06:06

df38f0a

Full Changelog: 2.0.0...2.0.1

Assets 2

VoxCPM 2 Released

a710128 released this 06 Apr 14:51

a010d62

修复lora/ft测试代码处传参错误的内容 by @Ayin1412 in #98
perf: optimize dataset length calculation via batch column access by @jayll1303 in #128
Add torchcodec to dependencies by @lrjerryli in #139
Print debug messages to stderr instead of stdout by @vytskalt in #141
Fix README's feature checkboxes by @zanellig in #147
Improve CLI argument validation and help text by @s3ldc in #161
Update train_voxcpm_finetune.py by @symhsym in #186
fix: use specific exceptions instead of bare except by @haosenwang1018 in #188
Add VoxCPM 2.