no render mac
parent
12dbf12299
commit
fffea2ddf0
|
|
@ -1,3 +1,3 @@
|
|||
[submodule "rust-whisper.d/gitea-whisper-rs"]
|
||||
[submodule "gitea-whisper-rs"]
|
||||
path = gitea-whisper-rs
|
||||
url = https://gitea.inhome.blapointe.com/bel/whisper-rs.git
|
||||
|
|
|
|||
|
|
@ -1,236 +1,284 @@
|
|||
use rust_whisper_lib;
|
||||
use rust_whisper_baked_lib;
|
||||
use clap::Parser;
|
||||
use listen_lib;
|
||||
use rust_whisper_baked_lib;
|
||||
use rust_whisper_lib;
|
||||
use std::thread;
|
||||
|
||||
fn main() {
|
||||
let flags = rust_whisper_lib::Flags::parse();
|
||||
match flags.wav.clone() {
|
||||
Some(_) => wav_channel(flags),
|
||||
None => channel(flags),
|
||||
};
|
||||
let flags = rust_whisper_lib::Flags::parse();
|
||||
match flags.wav.clone() {
|
||||
Some(_) => wav_channel(flags),
|
||||
None => channel(flags),
|
||||
};
|
||||
}
|
||||
|
||||
fn wav_channel(flags: rust_whisper_lib::Flags) {
|
||||
let mut w = new_destutterer();
|
||||
rust_whisper_baked_lib::wav_channel(
|
||||
flags.clone(),
|
||||
move |result: Result<rust_whisper_lib::Transcribed, String>| {
|
||||
match result {
|
||||
Ok(transcribed) => {
|
||||
let s = w.step(transcribed.to_string());
|
||||
println!("{}", s);
|
||||
},
|
||||
Err(msg) => { eprintln!("error: {}", msg); },
|
||||
};
|
||||
},
|
||||
);
|
||||
let mut w = new_destutterer();
|
||||
rust_whisper_baked_lib::wav_channel(
|
||||
flags.clone(),
|
||||
move |result: Result<rust_whisper_lib::Transcribed, String>| {
|
||||
match result {
|
||||
Ok(transcribed) => {
|
||||
let s = w.step(transcribed.to_string());
|
||||
println!("{}", s);
|
||||
}
|
||||
Err(msg) => {
|
||||
eprintln!("error: {}", msg);
|
||||
}
|
||||
};
|
||||
},
|
||||
);
|
||||
}
|
||||
|
||||
fn wav(flags: rust_whisper_lib::Flags, _path: String) {
|
||||
let mut w = new_destutterer();
|
||||
rust_whisper_baked_lib::wav(flags,
|
||||
move |result: Result<rust_whisper_lib::Transcribed, String>| {
|
||||
match result {
|
||||
Ok(transcribed) => {
|
||||
let s = w.step(transcribed.to_string());
|
||||
println!("{}", s);
|
||||
},
|
||||
Err(msg) => { eprintln!("error: {}", msg); },
|
||||
};
|
||||
},
|
||||
);
|
||||
let mut w = new_destutterer();
|
||||
rust_whisper_baked_lib::wav(
|
||||
flags,
|
||||
move |result: Result<rust_whisper_lib::Transcribed, String>| {
|
||||
match result {
|
||||
Ok(transcribed) => {
|
||||
let s = w.step(transcribed.to_string());
|
||||
println!("{}", s);
|
||||
}
|
||||
Err(msg) => {
|
||||
eprintln!("error: {}", msg);
|
||||
}
|
||||
};
|
||||
},
|
||||
);
|
||||
}
|
||||
|
||||
fn channel(flags: rust_whisper_lib::Flags) {
|
||||
let (send, recv) = std::sync::mpsc::sync_channel(100);
|
||||
let (send, recv) = std::sync::mpsc::sync_channel(100);
|
||||
|
||||
eprintln!("rust whisper baked lib channel...");
|
||||
thread::spawn(move || {
|
||||
let mut w = new_destutterer();
|
||||
rust_whisper_baked_lib::channel(
|
||||
flags.clone(),
|
||||
move |result: Result<rust_whisper_lib::Transcribed, String>| {
|
||||
match result {
|
||||
Ok(transcribed) => {
|
||||
let s = w.step(transcribed.to_string());
|
||||
println!("{}", s);
|
||||
},
|
||||
Err(msg) => { eprintln!("error: {}", msg); },
|
||||
};
|
||||
},
|
||||
recv,
|
||||
);
|
||||
});
|
||||
eprintln!("rust whisper baked lib channel...");
|
||||
thread::spawn(move || {
|
||||
let mut w = new_destutterer();
|
||||
rust_whisper_baked_lib::channel(
|
||||
flags.clone(),
|
||||
move |result: Result<rust_whisper_lib::Transcribed, String>| {
|
||||
match result {
|
||||
Ok(transcribed) => {
|
||||
let s = w.step(transcribed.to_string());
|
||||
println!("{}", s);
|
||||
}
|
||||
Err(msg) => {
|
||||
eprintln!("error: {}", msg);
|
||||
}
|
||||
};
|
||||
},
|
||||
recv,
|
||||
);
|
||||
});
|
||||
|
||||
eprintln!("listen lib main...");
|
||||
let flags = rust_whisper_lib::Flags::parse();
|
||||
match flags.stream_device {
|
||||
Some(device_name) => {
|
||||
eprintln!("with device ({}) '{}'", device_name.len(), &device_name);
|
||||
if device_name.len() == 0 {
|
||||
let mut i = 0;
|
||||
for device in listen_lib::devices() {
|
||||
eprintln!("[{}] {}", i, device);
|
||||
i += 1;
|
||||
eprintln!("listen lib main...");
|
||||
let flags = rust_whisper_lib::Flags::parse();
|
||||
match flags.stream_device {
|
||||
Some(device_name) => {
|
||||
eprintln!("with device ({}) '{}'", device_name.len(), &device_name);
|
||||
if device_name.len() == 0 {
|
||||
let mut i = 0;
|
||||
for device in listen_lib::devices() {
|
||||
eprintln!("[{}] {}", i, device);
|
||||
i += 1;
|
||||
}
|
||||
eprintln!("found {} devices", i);
|
||||
} else {
|
||||
listen_lib::main_with(
|
||||
|data| {
|
||||
send.send(data).unwrap();
|
||||
},
|
||||
device_name,
|
||||
);
|
||||
}
|
||||
eprintln!("found {} devices", i);
|
||||
} else {
|
||||
listen_lib::main_with(|data| {
|
||||
send.send(data).unwrap();
|
||||
}, device_name);
|
||||
}
|
||||
},
|
||||
None => {
|
||||
eprintln!("without any device");
|
||||
listen_lib::main(|data| {
|
||||
send.send(data).unwrap();
|
||||
});
|
||||
}
|
||||
}
|
||||
eprintln!("/listen lib main...");
|
||||
}
|
||||
None => {
|
||||
eprintln!("without any device");
|
||||
listen_lib::main(|data| {
|
||||
send.send(data).unwrap();
|
||||
});
|
||||
}
|
||||
}
|
||||
eprintln!("/listen lib main...");
|
||||
}
|
||||
|
||||
struct Destutterer {
|
||||
prev: Words,
|
||||
prev: Words,
|
||||
}
|
||||
|
||||
fn new_destutterer() -> Destutterer {
|
||||
Destutterer{prev: new_words()}
|
||||
Destutterer { prev: new_words() }
|
||||
}
|
||||
|
||||
impl Destutterer {
|
||||
fn step(&mut self, next: String) -> String {
|
||||
if next.len() == 0 {
|
||||
return next;
|
||||
}
|
||||
fn step(&mut self, next: String) -> String {
|
||||
if next.len() == 0 {
|
||||
return next;
|
||||
}
|
||||
|
||||
let next_words = Words::from_string(next.clone());
|
||||
let mut n = self.prev.comparable_len().clamp(0, next_words.comparable_len());
|
||||
//println!("n={} prev='{:?}' next='{:?}'", n, self.prev.to_comparable_words(), next_words.to_comparable_words());
|
||||
while n > 0 {
|
||||
let (prev_s, _) = self.prev.last_n_comparable_to_string(n);
|
||||
let (next_s, next_idx) = next_words.first_n_comparable_to_string(n);
|
||||
if prev_s == next_s {
|
||||
self.prev = next_words;
|
||||
return self.prev.skip(next_idx+1).to_string();
|
||||
}
|
||||
n -= 1;
|
||||
}
|
||||
self.prev = next_words;
|
||||
self.prev.to_string()
|
||||
}
|
||||
let next_words = Words::from_string(next.clone());
|
||||
let mut n = self
|
||||
.prev
|
||||
.comparable_len()
|
||||
.clamp(0, next_words.comparable_len());
|
||||
//println!("n={} prev='{:?}' next='{:?}'", n, self.prev.to_comparable_words(), next_words.to_comparable_words());
|
||||
while n > 0 {
|
||||
let (prev_s, _) = self.prev.last_n_comparable_to_string(n);
|
||||
let (next_s, next_idx) = next_words.first_n_comparable_to_string(n);
|
||||
if prev_s == next_s {
|
||||
self.prev = next_words;
|
||||
return self.prev.skip(next_idx + 1).to_string();
|
||||
}
|
||||
n -= 1;
|
||||
}
|
||||
self.prev = next_words;
|
||||
self.prev.to_string()
|
||||
}
|
||||
}
|
||||
|
||||
#[derive(Clone, Debug)]
|
||||
struct Words {
|
||||
raw: Vec<String>,
|
||||
raw: Vec<String>,
|
||||
}
|
||||
|
||||
fn new_words() -> Words {
|
||||
Words{raw: vec![]}
|
||||
Words { raw: vec![] }
|
||||
}
|
||||
|
||||
impl Words {
|
||||
fn from_string(s: String) -> Words {
|
||||
let mut result = Words{raw: vec![]};
|
||||
for word in s.split(" ") {
|
||||
let word = word.trim();
|
||||
if word.len() > 0 {
|
||||
result.raw.push(word.to_string());
|
||||
}
|
||||
}
|
||||
result
|
||||
}
|
||||
fn from_string(s: String) -> Words {
|
||||
let mut result = Words { raw: vec![] };
|
||||
for word in s.split(" ") {
|
||||
let word = word.trim();
|
||||
if word.len() > 0 {
|
||||
result.raw.push(word.to_string());
|
||||
}
|
||||
}
|
||||
result
|
||||
}
|
||||
|
||||
fn skip(&self, n: usize) -> Words {
|
||||
Words{
|
||||
raw: self.raw.iter().skip(n).map(|x| x.clone()).collect(),
|
||||
}
|
||||
}
|
||||
fn skip(&self, n: usize) -> Words {
|
||||
Words {
|
||||
raw: self.raw.iter().skip(n).map(|x| x.clone()).collect(),
|
||||
}
|
||||
}
|
||||
|
||||
fn last_n_comparable_to_string(&self, n: usize) -> (String, usize) {
|
||||
let v = self.to_comparable_words();
|
||||
let v = v[(v.len() - n).clamp(0, v.len())..].to_vec();
|
||||
return (v.iter().map(|x| x.s.clone().unwrap()).collect::<Vec<String>>().join(" "), v[0].idx)
|
||||
}
|
||||
fn last_n_comparable_to_string(&self, n: usize) -> (String, usize) {
|
||||
let v = self.to_comparable_words();
|
||||
let v = v[(v.len() - n).clamp(0, v.len())..].to_vec();
|
||||
return (
|
||||
v.iter()
|
||||
.map(|x| x.s.clone().unwrap())
|
||||
.collect::<Vec<String>>()
|
||||
.join(" "),
|
||||
v[0].idx,
|
||||
);
|
||||
}
|
||||
|
||||
fn first_n_comparable_to_string(&self, n: usize) -> (String, usize){
|
||||
let v = self.to_comparable_words();
|
||||
let v = v[0..n.clamp(0, v.len())].to_vec();
|
||||
return (v.iter().map(|x| x.s.clone().unwrap()).collect::<Vec<String>>().join(" "), v[v.len()-1].idx)
|
||||
}
|
||||
fn first_n_comparable_to_string(&self, n: usize) -> (String, usize) {
|
||||
let v = self.to_comparable_words();
|
||||
let v = v[0..n.clamp(0, v.len())].to_vec();
|
||||
return (
|
||||
v.iter()
|
||||
.map(|x| x.s.clone().unwrap())
|
||||
.collect::<Vec<String>>()
|
||||
.join(" "),
|
||||
v[v.len() - 1].idx,
|
||||
);
|
||||
}
|
||||
|
||||
fn comparable_len(&self) -> usize {
|
||||
self.to_comparable_words().len()
|
||||
}
|
||||
fn comparable_len(&self) -> usize {
|
||||
self.to_comparable_words().len()
|
||||
}
|
||||
|
||||
fn to_comparable_words(&self) -> Vec<Word> {
|
||||
self.to_words().iter()
|
||||
.filter(|x| x.s.is_some())
|
||||
.map(|x| x.clone())
|
||||
.collect()
|
||||
}
|
||||
fn to_comparable_words(&self) -> Vec<Word> {
|
||||
self.to_words()
|
||||
.iter()
|
||||
.filter(|x| x.s.is_some())
|
||||
.map(|x| x.clone())
|
||||
.collect()
|
||||
}
|
||||
|
||||
fn to_words(&self) -> Vec<Word> {
|
||||
let skips = stop_words::get("en");
|
||||
let stemmer = rust_stemmers::Stemmer::create(rust_stemmers::Algorithm::English);
|
||||
let strs = self.raw.iter()
|
||||
.map(|w| w.to_lowercase())
|
||||
.map(|w| w.chars().filter(|c| c.is_ascii_alphanumeric()).collect::<String>())
|
||||
.map(|w| stemmer.stem(&w).into_owned())
|
||||
.collect::<Vec<String>>();
|
||||
let mut result = vec![];
|
||||
for i in 0..strs.len() {
|
||||
result.push(Word{
|
||||
s: match skips.contains(&strs[i]) {
|
||||
true => None,
|
||||
false => Some(strs[i].clone()),
|
||||
},
|
||||
idx: i as usize,
|
||||
});
|
||||
}
|
||||
result
|
||||
}
|
||||
fn to_words(&self) -> Vec<Word> {
|
||||
let skips = stop_words::get("en");
|
||||
let stemmer = rust_stemmers::Stemmer::create(rust_stemmers::Algorithm::English);
|
||||
let strs = self
|
||||
.raw
|
||||
.iter()
|
||||
.map(|w| w.to_lowercase())
|
||||
.map(|w| {
|
||||
w.chars()
|
||||
.filter(|c| c.is_ascii_alphanumeric())
|
||||
.collect::<String>()
|
||||
})
|
||||
.map(|w| stemmer.stem(&w).into_owned())
|
||||
.collect::<Vec<String>>();
|
||||
let mut result = vec![];
|
||||
for i in 0..strs.len() {
|
||||
result.push(Word {
|
||||
s: match skips.contains(&strs[i]) {
|
||||
true => None,
|
||||
false => Some(strs[i].clone()),
|
||||
},
|
||||
idx: i as usize,
|
||||
});
|
||||
}
|
||||
result
|
||||
}
|
||||
|
||||
fn to_string(&self) -> String {
|
||||
self.raw.iter()
|
||||
.map(|x| x.clone())
|
||||
.collect::<Vec<String>>()
|
||||
.join(" ")
|
||||
}
|
||||
fn to_string(&self) -> String {
|
||||
self.raw
|
||||
.iter()
|
||||
.map(|x| x.clone())
|
||||
.collect::<Vec<String>>()
|
||||
.join(" ")
|
||||
}
|
||||
}
|
||||
|
||||
#[derive(Debug, Clone)]
|
||||
struct Word {
|
||||
s: Option<String>,
|
||||
idx: usize,
|
||||
s: Option<String>,
|
||||
idx: usize,
|
||||
}
|
||||
|
||||
#[cfg(test)]
|
||||
mod tests {
|
||||
use super::*;
|
||||
use super::*;
|
||||
|
||||
#[test]
|
||||
fn test_destutterer_stop_words() {
|
||||
let mut w = new_destutterer();
|
||||
assert_eq!("welcome to the internet".to_string(), w.step("welcome to the internet".to_string()));
|
||||
assert_eq!("have a look around".to_string(), w.step("welcome to the a internet; have a look around".to_string()));
|
||||
}
|
||||
#[test]
|
||||
fn test_destutterer_stop_words() {
|
||||
let mut w = new_destutterer();
|
||||
assert_eq!(
|
||||
"welcome to the internet".to_string(),
|
||||
w.step("welcome to the internet".to_string())
|
||||
);
|
||||
assert_eq!(
|
||||
"have a look around".to_string(),
|
||||
w.step("welcome to the a internet; have a look around".to_string())
|
||||
);
|
||||
}
|
||||
|
||||
#[test]
|
||||
fn test_destutterer_punctuation() {
|
||||
let mut w = new_destutterer();
|
||||
assert_eq!("cat, dog. cow? moose!".to_string(), w.step("cat, dog. cow? moose!".to_string()));
|
||||
assert_eq!("elephant! fez gator".to_string(), w.step("moose, elephant! fez gator".to_string()));
|
||||
assert_eq!("hij".to_string(), w.step("fez gator hij".to_string()));
|
||||
}
|
||||
#[test]
|
||||
fn test_destutterer_punctuation() {
|
||||
let mut w = new_destutterer();
|
||||
assert_eq!(
|
||||
"cat, dog. cow? moose!".to_string(),
|
||||
w.step("cat, dog. cow? moose!".to_string())
|
||||
);
|
||||
assert_eq!(
|
||||
"elephant! fez gator".to_string(),
|
||||
w.step("moose, elephant! fez gator".to_string())
|
||||
);
|
||||
assert_eq!("hij".to_string(), w.step("fez gator hij".to_string()));
|
||||
}
|
||||
|
||||
#[test]
|
||||
fn test_destutterer_basic() {
|
||||
let mut w = new_destutterer();
|
||||
assert_eq!("cat dog cow".to_string(), w.step(" cat dog cow ".to_string()));
|
||||
assert_eq!("moose".to_string(), w.step(" dog cow moose ".to_string()));
|
||||
}
|
||||
#[test]
|
||||
fn test_destutterer_basic() {
|
||||
let mut w = new_destutterer();
|
||||
assert_eq!(
|
||||
"cat dog cow".to_string(),
|
||||
w.step(" cat dog cow ".to_string())
|
||||
);
|
||||
assert_eq!("moose".to_string(), w.step(" dog cow moose ".to_string()));
|
||||
}
|
||||
}
|
||||
|
|
|
|||
Loading…
Reference in New Issue