oop

rust-whisper-baked-lib/src/lib.rs

@@ -33,7 +33,7 @@ pub fn f32_from_wav_file(path: &String) -> Result<Vec<f32>, String> {
 }
 
 fn get_fast() -> Vec<u8> {
-   include_bytes!("../../models/ggml-base.en.bin").to_vec()
+   include_bytes!("../../models/ggml-small.en.bin").to_vec()
 }
 
 fn get_good() -> Vec<u8> {

rust-whisper-baked/src/main.rs

@@ -48,11 +48,15 @@ fn channel(flags: rust_whisper_lib::Flags) {
 
    eprintln!("rust whisper baked lib channel...");
    thread::spawn(move || {
+      let mut w = new_destutterer();
       rust_whisper_baked_lib::channel(
          flags.clone(),
-         |result: Result<rust_whisper_lib::Transcribed, String>| {
+         move |result: Result<rust_whisper_lib::Transcribed, String>| {
             match result {
-               Ok(transcribed) => { println!("{}", transcribed.to_string()); },
+               Ok(transcribed) => {
+                  let s = w.step(transcribed.to_string());
+                  println!("{}", s);
+               },
                Err(msg) => { eprintln!("error: {}", msg); },
             };
          },
@@ -98,7 +102,7 @@ impl Destutterer {
       }
 
       let next_words = Words::from_string(next.clone());
-      let mut n = self.prev.to_comparable_words().len().clamp(0, next_words.to_comparable_words().len());
+      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);

todo.yaml

@@ -1,4 +1,5 @@
 todo:
+- whisper trims outside silence so head and tail never get hit
 - split on silence-ish instead of duration
 - rust-whisper warn when transcription time ~ input time
 scheduled: []