huggingface / huggingface/candle

Problems with Parler-TTS Mini Multilingual v1.1

Open
#2,709 0 comments 0 reactions 0 assignees View on GitHub
Dominant language
Rust
Stars
21k
Forks
1.8k
Avg merge
16h 42m
Merged PRs (30d)
25

Description

Hi, i love the library and the work your doing!

Recently tried the parler-tts example, works fine for the large-v1 model.

I wanted to try the mini multilingual one but get the following error:

```bash
avx: false, neon: true, simd128: false, f16c: false
temp: 0.00 repeat-penalty: 1.00 repeat-last-n: 64
retrieved the files in 54.020418s
Error: data did not match any variant of untagged enum ModelWrapper at line 699311 column 1
```

Added this modifications to the example:

```rust
let model_id = match args.model_id {
Some(model_id) => model_id.to_string(),
None => match args.which {
Which::LargeV1 => "parler-tts/parler-tts-large-v1".to_string(),
Which::MiniV1 => "parler-tts/parler-tts-mini-v1".to_string(),
Which::Multi => "parler-tts/parler-tts-mini-multilingual-v1.1".to_string(), // added this line
},
};
```

```rust
let model_files = match args.model_file {
Some(m) => vec![m.into()],
None => match args.which {
Which::MiniV1 => vec![repo.get("model.safetensors")?],
Which::Multi => vec![repo.get("model.safetensors")?],
Which::LargeV1 => {
candle_examples::hub_load_safetensors(&repo, "model.safetensors.index.json")?
}
},
};
```

```rust
#[derive(Clone, Debug, Copy, PartialEq, Eq, clap::ValueEnum)]
enum Which {
#[value(name = "large-v1")]
LargeV1,
#[value(name = "mini-v1")]
MiniV1,
#[value(name = "mini-multi")]
Multi,
}
```

And called the tool with
`--which mini-multi`

The download of the `model.safetensors` has worked but then i get an error.

I'am running on an mac with the `--features metal` parameter

Do i have missed something?

Contributor guide

No contributing guide indexed for this repository

Research direction

Start with the parler-tts example and its model-loading path, using the reported model ID and model.safetensors file as the reproduction. Compare how the multilingual model is loaded with the working large-v1 model and run the example with --which mini-multi and --features metal. Done means the multilingual model loads without the ModelWrapper error or the incompatibility is clearly documented.

Written by the indexing model from the issue text.

Assessment

Tech stack
rust
Domain
machine-learning
Issue type
Bug
Difficulty
3/5
Estimated time
1-2 days
Activity status
Stale
Clarity
Mostly clear
Newbie friendliness
45/100

Get new issues in your inbox

A short digest of beginner-friendly GitHub issues.