diff --git a/src/backends/speech_dispatcher.rs b/src/backends/speech_dispatcher.rs
index 009a8fb..5717c61 100644
--- a/src/backends/speech_dispatcher.rs
+++ b/src/backends/speech_dispatcher.rs
@@ -6,7 +6,7 @@ use lazy_static::*;
 use log::{info, trace};
 use speech_dispatcher::*;
 
-use crate::{Backend, Error, Features};
+use crate::{Backend, Error, Features, UtteranceId};
 
 pub struct SpeechDispatcher(Connection);
 
@@ -59,7 +59,7 @@ impl Backend for SpeechDispatcher {
         }
     }
 
-    fn speak(&mut self, text: &str, interrupt: bool) -> Result<(), Error> {
+    fn speak(&mut self, text: &str, interrupt: bool) -> Result<Option<UtteranceId>, Error> {
         trace!("speak({}, {})", text, interrupt);
         if interrupt {
             self.stop()?;
@@ -68,11 +68,15 @@ impl Backend for SpeechDispatcher {
         if single_char {
             self.0.set_punctuation(Punctuation::All);
         }
-        self.0.say(Priority::Important, text);
+        let id = self.0.say(Priority::Important, text);
         if single_char {
             self.0.set_punctuation(Punctuation::None);
         }
-        Ok(())
+        if let Some(id) = id {
+            Ok(Some(UtteranceId::SpeechDispatcher(id)))
+        } else {
+            Err(Error::NoneError)
+        }
     }
 
     fn stop(&mut self) -> Result<(), Error> {
diff --git a/src/backends/tolk.rs b/src/backends/tolk.rs
index 370da65..28c3f42 100644
--- a/src/backends/tolk.rs
+++ b/src/backends/tolk.rs
@@ -2,7 +2,7 @@
 use log::{info, trace};
 use tolk::Tolk as TolkPtr;
 
-use crate::{Backend, Error, Features};
+use crate::{Backend, Error, Features, UtteranceId};
 
 pub struct Tolk(TolkPtr);
 
@@ -26,10 +26,10 @@ impl Backend for Tolk {
         }
     }
 
-    fn speak(&mut self, text: &str, interrupt: bool) -> Result<(), Error> {
+    fn speak(&mut self, text: &str, interrupt: bool) -> Result<Option<UtteranceId>, Error> {
         trace!("speak({}, {})", text, interrupt);
         self.0.speak(text, interrupt);
-        Ok(())
+        Ok(None)
     }
 
     fn stop(&mut self) -> Result<(), Error> {
diff --git a/src/backends/web.rs b/src/backends/web.rs
index 664793c..89db8a9 100644
--- a/src/backends/web.rs
+++ b/src/backends/web.rs
@@ -1,8 +1,11 @@
 #[cfg(target_arch = "wasm32")]
+use std::sync::Mutex;
+
+use lazy_static::lazy_static;
 use log::{info, trace};
 use web_sys::SpeechSynthesisUtterance;
 
-use crate::{Backend, Error, Features};
+use crate::{Backend, Error, Features, UtteranceId};
 
 pub struct Web {
     rate: f32,
@@ -10,6 +13,10 @@ pub struct Web {
     volume: f32,
 }
 
+lazy_static! {
+    static ref NEXT_UTTERANCE_ID: Mutex<u64> = Mutex::new(0);
+}
+
 impl Web {
     pub fn new() -> Result<Self, Error> {
         info!("Initializing Web backend");
@@ -32,7 +39,7 @@ impl Backend for Web {
         }
     }
 
-    fn speak(&mut self, text: &str, interrupt: bool) -> Result<(), Error> {
+    fn speak(&mut self, text: &str, interrupt: bool) -> Result<Option<UtteranceId>, Error> {
         trace!("speak({}, {})", text, interrupt);
         let utterance = SpeechSynthesisUtterance::new_with_text(text).unwrap();
         utterance.set_rate(self.rate);
@@ -44,8 +51,12 @@ impl Backend for Web {
         if let Some(window) = web_sys::window() {
             let speech_synthesis = window.speech_synthesis().unwrap();
             speech_synthesis.speak(&utterance);
+            let mut utterance_id = NEXT_UTTERANCE_ID.lock().unwrap();
+            *utterance_id += 1;
+            Ok(Some(UtteranceId::Web(*utterance_id)))
+        } else {
+            Err(Error::NoneError)
         }
-        Ok(())
     }
 
     fn stop(&mut self) -> Result<(), Error> {
diff --git a/src/backends/winrt.rs b/src/backends/winrt.rs
index 7464f58..32fb219 100644
--- a/src/backends/winrt.rs
+++ b/src/backends/winrt.rs
@@ -1,4 +1,7 @@
 #[cfg(windows)]
+use std::sync::Mutex;
+
+use lazy_static::lazy_static;
 use log::{info, trace};
 
 use tts_winrt_bindings::windows::media::core::MediaSource;
@@ -7,7 +10,7 @@ use tts_winrt_bindings::windows::media::playback::{
 };
 use tts_winrt_bindings::windows::media::speech_synthesis::SpeechSynthesizer;
 
-use crate::{Backend, Error, Features};
+use crate::{Backend, Error, Features, UtteranceId};
 
 impl From<winrt::Error> for Error {
     fn from(e: winrt::Error) -> Self {
@@ -21,6 +24,10 @@ pub struct WinRT {
     playback_list: MediaPlaybackList,
 }
 
+lazy_static! {
+    static ref NEXT_UTTERANCE_ID: Mutex<u64> = Mutex::new(0);
+}
+
 impl WinRT {
     pub fn new() -> std::result::Result<Self, Error> {
         info!("Initializing WinRT backend");
@@ -55,7 +62,11 @@ impl Backend for WinRT {
         }
     }
 
-    fn speak(&mut self, text: &str, interrupt: bool) -> std::result::Result<(), Error> {
+    fn speak(
+        &mut self,
+        text: &str,
+        interrupt: bool,
+    ) -> std::result::Result<Option<UtteranceId>, Error> {
         trace!("speak({}, {})", text, interrupt);
         if interrupt {
             self.stop()?;
@@ -76,7 +87,9 @@ impl Backend for WinRT {
         if !self.is_speaking()? {
             self.player.play()?;
         }
-        Ok(())
+        let mut utterance_id = NEXT_UTTERANCE_ID.lock().unwrap();
+        *utterance_id += 1;
+        Ok(Some(UtteranceId::WinRT(*utterance_id)))
     }
 
     fn stop(&mut self) -> std::result::Result<(), Error> {
diff --git a/src/lib.rs b/src/lib.rs
index 56b9f32..2d494cf 100644
--- a/src/lib.rs
+++ b/src/lib.rs
@@ -40,6 +40,16 @@ pub enum Backends {
     AvFoundation,
 }
 
+#[derive(Clone, Copy, Debug, PartialEq)]
+pub enum UtteranceId {
+    #[cfg(target_os = "linux")]
+    SpeechDispatcher(i32),
+    #[cfg(target_arch = "wasm32")]
+    Web(u64),
+    #[cfg(windows)]
+    WinRT(u64),
+}
+
 pub struct Features {
     pub stop: bool,
     pub rate: bool,
@@ -80,7 +90,7 @@ pub enum Error {
 
 pub trait Backend {
     fn supported_features(&self) -> Features;
-    fn speak(&mut self, text: &str, interrupt: bool) -> Result<(), Error>;
+    fn speak(&mut self, text: &str, interrupt: bool) -> Result<Option<UtteranceId>, Error>;
     fn stop(&mut self) -> Result<(), Error>;
     fn min_rate(&self) -> f32;
     fn max_rate(&self) -> f32;
@@ -184,9 +194,12 @@ impl TTS {
     /**
      * Speaks the specified text, optionally interrupting current speech.
      */
-    pub fn speak<S: Into<String>>(&mut self, text: S, interrupt: bool) -> Result<&Self, Error> {
-        self.0.speak(text.into().as_str(), interrupt)?;
-        Ok(self)
+    pub fn speak<S: Into<String>>(
+        &mut self,
+        text: S,
+        interrupt: bool,
+    ) -> Result<Option<UtteranceId>, Error> {
+        self.0.speak(text.into().as_str(), interrupt)
     }
 
     /**