wavo 0.3.1

Speech-to-text in pure Rust on the GPU (Metal, Vulkan, DX12): Parakeet, GigaAM, Whisper and Moonshine, as a library and an ollama-like CLI.
Documentation 
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85

// Copyright (c) vladkens | MIT License | https://github.com/vladkens/wavo

/// Everything that can fail in wavo: loading a model, running it on the GPU and, with the `hub`
/// feature, managing downloads.
///
/// It is `#[non_exhaustive]` because the `hub` feature adds variants, and features unify across a
/// build: keep a `_` arm when matching on it.
#[derive(Debug, thiserror::Error)]
#[non_exhaustive]
pub enum Error {
  /// Reading the model file, or the cache's files and folders with the `hub` feature, failed.
  #[error(transparent)]
  Io(#[from] std::io::Error),
  /// The file is not a model wavo supports: not a GGUF, an architecture other than `gigaam`,
  /// `parakeet`, `whisper` and `moonshine`, or a missing or mismatched tensor or metadata key.
  #[error("invalid model: {0}")]
  Model(String),
  /// No GPU adapter (Metal, Vulkan, or DX12 on Windows) is available.
  #[error("no GPU adapter: {0}")]
  Adapter(#[from] wgpu::RequestAdapterError),
  /// The GPU adapter refused a device with the limits the model needs.
  #[error("GPU device: {0}")]
  Device(#[from] wgpu::RequestDeviceError),
  /// Reading results back from the GPU failed.
  #[error("GPU readback: {0}")]
  Readback(#[from] wgpu::BufferAsyncError),
  /// Mapping a GPU buffer for reading failed.
  #[error("GPU readback: {0}")]
  Map(#[from] wgpu::MapRangeError),
  /// Waiting for the GPU failed, for example after a device loss.
  #[error("GPU poll: {0}")]
  Poll(#[from] wgpu::PollError),
  /// An error the GPU reported during a call (out of memory, a lost device), or work the driver
  /// dropped unfinished: some drivers cancel GPU work that runs for more than about 20 s.
  #[error("GPU: {0}")]
  Gpu(String),
  /// A name that is neither a short nor a full name in [`hub::MODELS`](crate::hub::MODELS).
  #[cfg(feature = "hub")]
  #[cfg_attr(docsrs, doc(cfg(feature = "hub")))]
  #[error("unknown model {name:?}{} Known models: {}", crate::hub::hint(*suggestion), crate::hub::names())]
  UnknownModel {
    /// The name as given.
    name: String,
    /// The closest known name, when one is a likely typo of `name`.
    suggestion: Option<&'static str>,
  },
  /// A request to the Hugging Face Hub failed, including a download that broke off midway.
  /// [`Hub::pull`](crate::hub::Hub::pull) has already retried what may pass on another try.
  #[cfg(feature = "hub")]
  #[cfg_attr(docsrs, doc(cfg(feature = "hub")))]
  #[error("{url}: {message}")]
  Http {
    /// The URL requested.
    url: String,
    /// The HTTP status of an error response (404: no such file); `None` when no response came
    /// or the body broke off.
    status: Option<u16>,
    /// What went wrong.
    message: String,
  },
  /// The Hub's answer is not as expected: a response without the commit, ETag or size headers,
  /// or a download of the wrong size.
  #[cfg(feature = "hub")]
  #[cfg_attr(docsrs, doc(cfg(feature = "hub")))]
  #[error("{0}")]
  Hub(String),
  /// The progress callback of [`Hub::pull`](crate::hub::Hub::pull) returned
  /// [`ControlFlow::Break`](std::ops::ControlFlow::Break). The partial download stays in the
  /// cache, and the next pull of the model resumes it.
  #[cfg(feature = "hub")]
  #[cfg_attr(docsrs, doc(cfg(feature = "hub")))]
  #[error("download cancelled")]
  Cancelled,
}

/// `Result` with wavo's [`Error`].
pub type Result<T> = std::result::Result<T, Error>;

/// Shorthand for `Error::Model` with a formatted message.
macro_rules! bail {
  ($($arg:tt)*) => {
    return Err($crate::error::Error::Model(format!($($arg)*)))
  };
}
pub(crate) use bail;