voxr 0.2.0

Audio and speech: codecs, DSP, G2P, voices, and TTS/ASR pipelines over boostr models
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
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135

//! Sample-level audio DSP, file codecs, voices, and product pipelines.
//!
//! Placement rule: code that operates on `Tensor<R>` with backend kernels
//! belongs in `boostr::model::audio`. Code that operates on `Vec<f32>`
//! samples, files, directories, voices, or text pipelines belongs here.
//!
//! Always compiled: RIFF/WAVE in and out ([`wav`]), the polyphase
//! [`resample`](mod@resample)rs (general, ffmpeg-exact and SciPy-exact), the single-sound
//! [`metrics`] and the [`sound_review`] built on them, the reference-take
//! [`enhance`] chain, [`pitch`], [`quality`], two-signal [`spectral`]
//! distance, the `atempo`-compatible tempo [`stretch`], ASR [`eval`]
//! metrics, the signal and summary halves of the [`tts_eval`] quality gate,
//! and the [`Lang`] tag a voice carries. Feature gates add the
//! compressed-audio [`decode`]rs, the [`clap`] audio-text scorer, the
//! misaki G2P backend ([`g2p`]), the spaCy `en_core_web_sm` tokenizer and
//! tagger, the [`tts`] surface, the [`kokoro`] and [`voxcpm`]
//! engines, the [`whisper`] bundle, the [`vad`] segmenter, [`corpus`]
//! preparation, and the pace, intelligibility and timbre legs of
//! [`tts_eval`].
//!
//! [`sound_review`] implements the sound review of the motion-video skill
//! (<https://github.com/farhan-syah/motion-video-skill>). Docs here call it
//! "the motion-video skill" or "the skill".

pub mod enhance;
pub mod error;
pub mod eval;
pub mod ffmpeg;
pub mod g2p;
pub mod lang;
pub mod metrics;
pub mod pitch;
pub mod quality;
pub mod resample;
pub mod sound_review;
pub mod spectral;
pub mod stretch;
pub mod tts_eval;
pub mod wav;

#[cfg(feature = "clap")]
pub mod clap;
#[cfg(feature = "corpus")]
pub mod corpus;
#[cfg(feature = "decode")]
pub mod decode;
#[cfg(feature = "kokoro")]
pub mod kokoro;
#[cfg(feature = "tts")]
pub mod tts;
#[cfg(feature = "vad")]
pub mod vad;
#[cfg(feature = "voxcpm")]
pub mod voxcpm;
#[cfg(feature = "whisper")]
pub mod whisper;

pub use error::{Error, Result};

#[cfg(feature = "corpus")]
pub use corpus::{
    CorpusOptions, MAX_UTTERANCE_SECS, PRETRAINED_TOKENIZER_NAMES, SpeechCorpusBuilder,
    TextTokenizer, Utterance, check_max_speech_duration, pack_utterances,
    pack_utterances_with_layout,
};
#[cfg(feature = "decode")]
pub use decode::{
    decode_audio, decode_audio_file_mono_at, decode_audio_like_ffmpeg, decode_audio_mono_at,
    extension_hint,
};
pub use eval::{
    ErrorRate, align, by_group, canonical_character_error_rate, canonical_word_error_rate,
    canonicalize, character_error_rate, grand_total, normalize, total, word_error_rate,
};
pub use ffmpeg::{ffmpeg_program, ffprobe_program};
pub use lang::{Lang, LangTag};
pub use pitch::{PitchOptions, PitchTrack, estimate_pitch};
pub use quality::{TakeQuality, measure_quality};
pub use resample::{
    DEFAULT_TAPS_PER_PHASE, MAX_FILTER_TAPS, PolySample, ffmpeg_default_layout,
    ffmpeg_resampled_len, librosa_output_len, resample, resample_interleaved_like_ffmpeg,
    resample_like_ffmpeg, resample_like_librosa, resample_poly, resample_with_taps,
    to_mono_at_rate, to_mono_like_ffmpeg,
};
pub use spectral::{SpectralDistance, multi_resolution_stft_distance, snr_db};
pub use stretch::{MAX_TEMPO, MIN_TEMPO, PcmSample, atempo, atempo_interleaved, atempo_pcm16};
#[cfg(feature = "tts")]
pub use tts::{
    KOKORO_VOICES, SynthesizeOptions, TtsBundle, TtsEngine, TtsError, Voice, default_kokoro_voices,
};
#[cfg(feature = "voxcpm")]
pub use tts_eval::{F0Stats, TimbreProxy, TimbreScorer, timbre_proxy};
#[cfg(feature = "whisper")]
pub use tts_eval::{Intelligibility, IntelligibilityClient, IntelligibilityScorer};
pub use tts_eval::{
    MetricStat, RowScore, SignalStats, Summary, Thresholds, UNLABELLED_AXIS, signal_stats,
    violations,
};
#[cfg(feature = "vad")]
pub use tts_eval::{Pace, pace, pace_from_segments};
#[cfg(feature = "vad")]
pub use vad::{SpeechSegment, VadSegmentOptions, segments_from_probabilities, speech_timestamps};
pub use wav::{
    WavData, WavF64, decode_wav, decode_wav_f64, encode_f32_raw, encode_pcm16_raw, encode_wav_f32,
    encode_wav_pcm16, encode_wav_pcm16_values, to_mono, wav_stream_header_f32,
    wav_stream_header_pcm16,
};
#[cfg(feature = "whisper")]
pub use whisper::{
    TranscribeOptions, Transcription, WhisperBundle, WhisperGenerationConfig, WordTiming,
};

#[cfg(all(test, feature = "decode"))]
pub(crate) mod test_utils {
    use std::path::PathBuf;

    /// Resolve a real-audio fixture: `$AUDIO_CORPUS_FLAC`, else the first
    /// `.flac` in `$AUDIO_CORPUS_DIR` by sorted name. `None` when neither is
    /// set or the resolved path is absent, so callers skip.
    pub(crate) fn corpus_flac() -> Option<PathBuf> {
        if let Ok(p) = std::env::var("AUDIO_CORPUS_FLAC") {
            let path = PathBuf::from(p);
            return path.exists().then_some(path);
        }
        let dir = PathBuf::from(std::env::var("AUDIO_CORPUS_DIR").ok()?);
        let mut flacs: Vec<PathBuf> = std::fs::read_dir(dir)
            .ok()?
            .filter_map(|entry| entry.ok())
            .map(|entry| entry.path())
            .filter(|path| path.extension().is_some_and(|ext| ext == "flac"))
            .collect();
        flacs.sort();
        flacs.into_iter().next()
    }
}