add another example
This commit is contained in:
parent
a329e0fa58
commit
406fcae694
@ -1,5 +1,6 @@
|
||||
[workspace]
|
||||
members = ["sys"]
|
||||
exclude = ["examples/full_usage"]
|
||||
|
||||
[package]
|
||||
name = "whisper-rs"
|
||||
|
BIN
examples/full_usage/2830-3980-0043.wav
Normal file
BIN
examples/full_usage/2830-3980-0043.wav
Normal file
Binary file not shown.
10
examples/full_usage/Cargo.toml
Normal file
10
examples/full_usage/Cargo.toml
Normal file
@ -0,0 +1,10 @@
|
||||
[package]
|
||||
name = "full_usage"
|
||||
version = "0.1.0"
|
||||
edition = "2021"
|
||||
|
||||
# See more keys and their definitions at https://doc.rust-lang.org/cargo/reference/manifest.html
|
||||
|
||||
[dependencies]
|
||||
hound = "3"
|
||||
whisper-rs = { path = "../.." }
|
62
examples/full_usage/src/main.rs
Normal file
62
examples/full_usage/src/main.rs
Normal file
@ -0,0 +1,62 @@
|
||||
#![allow(clippy::uninlined_format_args)]
|
||||
|
||||
use hound::{SampleFormat, WavReader};
|
||||
use std::path::Path;
|
||||
use whisper_rs::{FullParams, SamplingStrategy, WhisperContext};
|
||||
|
||||
fn parse_wav_file(path: &Path) -> Vec<i16> {
|
||||
let reader = WavReader::open(path).expect("failed to read file");
|
||||
|
||||
if reader.spec().channels != 1 {
|
||||
panic!("expected mono audio file");
|
||||
}
|
||||
if reader.spec().sample_format != SampleFormat::Int {
|
||||
panic!("expected integer sample format");
|
||||
}
|
||||
if reader.spec().sample_rate != 16000 {
|
||||
panic!("expected 16KHz sample rate");
|
||||
}
|
||||
if reader.spec().bits_per_sample != 16 {
|
||||
panic!("expected 16 bits per sample");
|
||||
}
|
||||
|
||||
reader
|
||||
.into_samples::<i16>()
|
||||
.map(|x| x.expect("sample"))
|
||||
.collect::<Vec<_>>()
|
||||
}
|
||||
|
||||
fn main() {
|
||||
let arg1 = std::env::args()
|
||||
.nth(1)
|
||||
.expect("first argument should be path to WAV file");
|
||||
let audio_path = Path::new(&arg1);
|
||||
if !audio_path.exists() && !audio_path.is_file() {
|
||||
panic!("expected a file");
|
||||
}
|
||||
let arg2 = std::env::args()
|
||||
.nth(2)
|
||||
.expect("second argument should be path to Whisper model");
|
||||
let whisper_path = Path::new(&arg2);
|
||||
if !whisper_path.exists() && !whisper_path.is_file() {
|
||||
panic!("expected a whisper directory")
|
||||
}
|
||||
|
||||
let original_samples = parse_wav_file(audio_path);
|
||||
let samples = whisper_rs::convert_integer_to_float_audio(&original_samples);
|
||||
|
||||
let mut ctx =
|
||||
WhisperContext::new(&whisper_path.to_string_lossy()).expect("failed to open model");
|
||||
let params = FullParams::new(SamplingStrategy::default());
|
||||
|
||||
ctx.full(params, &samples)
|
||||
.expect("failed to convert samples");
|
||||
|
||||
let num_segments = ctx.full_n_segments();
|
||||
for i in 0..num_segments {
|
||||
let segment = ctx.full_get_segment_text(i).expect("failed to get segment");
|
||||
let start_timestamp = ctx.full_get_segment_t0(i);
|
||||
let end_timestamp = ctx.full_get_segment_t1(i);
|
||||
println!("[{} - {}]: {}", start_timestamp, end_timestamp, segment);
|
||||
}
|
||||
}
|
Loading…
Reference in New Issue
Block a user