baked-lib wav_channel to base

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

@@ -24,7 +24,7 @@ pub fn wav_channel<F>(
    handler_fn: F
 ) where F: FnMut(Result<rust_whisper_lib::Transcribed, String>) + Send + 'static {
    flags.model_path = None;
-   flags.model_buffer = Some(include_bytes!("../../models/ggml-distil-medium.en.bin").to_vec());
+   flags.model_buffer = Some(include_bytes!("../../models/ggml-base.en.bin").to_vec());
    rust_whisper_lib::wav_channel(flags, handler_fn);
 }
 

rust-whisper-baked/src/main.rs

@@ -13,11 +13,15 @@ fn main() {
 }
 
 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) => { println!("{}", transcribed.to_string()); },
+            Ok(transcribed) => {
+               let s = w.step(transcribed.to_string());
+               println!("{}", s);
+            },
             Err(msg) => { eprintln!("error: {}", msg); },
          };
       },
@@ -25,10 +29,14 @@ fn wav_channel(flags: rust_whisper_lib::Flags) {
 }
 
 fn wav(flags: rust_whisper_lib::Flags, _path: String) {
+   let mut w = new_destutterer();
    rust_whisper_baked_lib::wav(flags,
-      |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); },
          };
       },
@@ -74,3 +82,66 @@ fn channel(flags: rust_whisper_lib::Flags) {
    }
    eprintln!("/listen lib main...");
 }
+
+struct Destutterer {
+   prev: Option<String>,
+}
+
+fn new_destutterer() -> Destutterer {
+   Destutterer{prev: None}
+}
+
+impl Destutterer {
+   fn step(&mut self, next: String) -> String {
+      let next = next.trim().to_string();
+      if next.len() == 0 {
+         return next;
+      }
+      match &self.prev {
+         None => {
+            self.prev = Some(next.clone());
+            next
+         },
+         Some(prev) => {
+            let without_trailing_punctuation = {
+               let mut next = next.clone();
+               while next.ends_with("?") || next.ends_with(".") {
+                  next = next[..next.len()-1].to_string();
+               }
+               next
+            };
+            let trailing_punctuation = next[without_trailing_punctuation.len() ..].to_string();
+            let next = without_trailing_punctuation;
+            let next = {
+               let mut n = prev.len().clamp(0, next.len());
+               while n > 0 {
+                  if prev[prev.len() - n..] == next[..n] {
+                     break;
+                  }
+                  n -= 1;
+               }
+               next[n..].to_string()
+            };
+            if next.len() == 0 {
+               return "".to_string();
+            }
+            self.prev = Some(next.clone());
+            next + &trailing_punctuation
+         },
+      }
+   }
+}
+
+#[cfg(test)]
+mod tests {
+   use super::*;
+
+   #[test]
+   fn test_destutterer() {
+      let mut w = new_destutterer();
+      assert_eq!("abcde".to_string(), w.step("abcde".to_string()));
+      assert_eq!("fg".to_string(), w.step("cdefg".to_string()));
+      assert_eq!("hij".to_string(), w.step("fghij".to_string()));
+      assert_eq!("fghij".to_string(), w.step("fghij".to_string()));
+   }
+}

rust-whisper-lib/src/lib.rs

@@ -230,7 +230,9 @@ impl Impl {
       let result = whispered
          .after(&(self.stream_head * 100.0))
          .before(&(self.stream_tail * 100.0));
-      (self.handler_fn.as_mut().unwrap())(Ok(result));
+      if result.to_string().trim().len() > 0 {
+         (self.handler_fn.as_mut().unwrap())(Ok(result));
+      }
    }
 
    fn on_error(&mut self, msg: String) {