diff --git a/tycode-core/src/voice/stt/aws_transcribe.rs b/tycode-core/src/voice/stt/aws_transcribe.rs
index 6422e36..076298f 100644
--- a/tycode-core/src/voice/stt/aws_transcribe.rs
+++ b/tycode-core/src/voice/stt/aws_transcribe.rs
@@ -14,7 +14,7 @@ use aws_sdk_transcribestreaming::{
 use tokio::sync::mpsc;
 
 use super::provider::{AudioSink, SpeechToText, TranscriptionStream};
-use super::types::{Speaker, TranscriptionChunk};
+use super::types::{Speaker, TranscriptionChunk, TranscriptionError};
 use crate::voice::audio::AudioProfile;
 
 /// Configuration for AWS Transcribe streaming
@@ -86,7 +86,7 @@ impl SpeechToText for AwsTranscribe {
     }
 
     async fn start(&self) -> Result<(AudioSink, TranscriptionStream)> {
-        let (result_tx, result_rx) = mpsc::channel::<TranscriptionChunk>(100);
+        let (result_tx, result_rx) = mpsc::channel::<Result<TranscriptionChunk, TranscriptionError>>(100);
         let (audio_tx, mut audio_rx) = mpsc::channel::<Vec<u8>>(100);
 
         let language_code = Self::parse_language_code(&self.config.language_code);
@@ -116,20 +116,36 @@ impl SpeechToText for AwsTranscribe {
             let output = match response {
                 Ok(output) => output,
                 Err(e) => {
-                    tracing::error!("Failed to start AWS Transcribe stream: {e:?}");
+                    let error = TranscriptionError::StartupFailed {
+                        message: format!("{e:?}"),
+                    };
+                    let _ = result_tx.send(Err(error)).await;
                     return;
                 }
             };
 
             let mut transcript_stream = output.transcript_result_stream;
 
-            while let Ok(Some(event)) = transcript_stream.recv().await {
-                let TranscriptResultStream::TranscriptEvent(transcript_event) = event else {
-                    continue;
-                };
-
-                for chunk in extract_chunks(transcript_event) {
-                    if result_tx.send(chunk).await.is_err() {
+            loop {
+                match transcript_stream.recv().await {
+                    Ok(Some(event)) => {
+                        let TranscriptResultStream::TranscriptEvent(transcript_event) = event
+                        else {
+                            continue;
+                        };
+
+                        for chunk in extract_chunks(transcript_event) {
+                            if result_tx.send(Ok(chunk)).await.is_err() {
+                                return;
+                            }
+                        }
+                    }
+                    Ok(None) => break,
+                    Err(e) => {
+                        let error = TranscriptionError::StreamError {
+                            message: format!("{e:?}"),
+                        };
+                        let _ = result_tx.send(Err(error)).await;
                         return;
                     }
                 }
diff --git a/tycode-core/src/voice/stt/provider.rs b/tycode-core/src/voice/stt/provider.rs
index aae83fb..f025336 100644
--- a/tycode-core/src/voice/stt/provider.rs
+++ b/tycode-core/src/voice/stt/provider.rs
@@ -2,7 +2,7 @@ use anyhow::{Context, Result};
 use async_trait::async_trait;
 use tokio::sync::mpsc;
 
-use super::types::TranscriptionChunk;
+use super::types::{TranscriptionChunk, TranscriptionError};
 use crate::voice::audio::AudioProfile;
 
 /// Trait for speech-to-text providers
@@ -41,17 +41,17 @@ impl AudioSink {
 
 /// Handle for receiving transcription results
 pub struct TranscriptionStream {
-    receiver: mpsc::Receiver<TranscriptionChunk>,
+    receiver: mpsc::Receiver<Result<TranscriptionChunk, TranscriptionError>>,
 }
 
 impl TranscriptionStream {
-    pub fn new(receiver: mpsc::Receiver<TranscriptionChunk>) -> Self {
+    pub fn new(receiver: mpsc::Receiver<Result<TranscriptionChunk, TranscriptionError>>) -> Self {
         Self { receiver }
     }
 
-    /// Receive the next transcription chunk
+    /// Receive the next transcription result
     /// Returns None when the stream ends
-    pub async fn recv(&mut self) -> Option<TranscriptionChunk> {
+    pub async fn recv(&mut self) -> Option<Result<TranscriptionChunk, TranscriptionError>> {
         self.receiver.recv().await
     }
 }
diff --git a/tycode-core/src/voice/stt/types.rs b/tycode-core/src/voice/stt/types.rs
index 5b7620e..f448ed0 100644
--- a/tycode-core/src/voice/stt/types.rs
+++ b/tycode-core/src/voice/stt/types.rs
@@ -1,4 +1,25 @@
 use serde::{Deserialize, Serialize};
+use std::fmt;
+
+/// Errors that can occur during transcription
+#[derive(Debug, Clone)]
+pub enum TranscriptionError {
+    /// AWS Transcribe failed to start streaming
+    StartupFailed { message: String },
+    /// Stream error during transcription
+    StreamError { message: String },
+}
+
+impl fmt::Display for TranscriptionError {
+    fn fmt(&self, f: &mut fmt::Formatter<'_>) -> fmt::Result {
+        match self {
+            Self::StartupFailed { message } => write!(f, "Transcription startup failed: {message}"),
+            Self::StreamError { message } => write!(f, "Transcription stream error: {message}"),
+        }
+    }
+}
+
+impl std::error::Error for TranscriptionError {}
 
 /// A chunk of transcribed text
 #[derive(Debug, Clone, Serialize, Deserialize)]
diff --git a/tycode-core/tests/voice.rs b/tycode-core/tests/voice.rs
index 6ea83f2..e88a346 100644
--- a/tycode-core/tests/voice.rs
+++ b/tycode-core/tests/voice.rs
@@ -133,7 +133,7 @@ async fn test_aws_transcribe_from_file() {
     while tokio::time::Instant::now() < deadline {
         match tokio::time::timeout(tokio::time::Duration::from_secs(5), transcriptions.recv()).await
         {
-            Ok(Some(chunk)) => {
+            Ok(Some(Ok(chunk))) => {
                 println!(
                     "Received: {} (partial: {}, speaker: {:?})",
                     chunk.text, chunk.is_partial, chunk.speaker
@@ -142,6 +142,10 @@ async fn test_aws_transcribe_from_file() {
                     results.push(chunk.text);
                 }
             }
+            Ok(Some(Err(e))) => {
+                println!("Transcription error: {}", e);
+                break;
+            }
             Ok(None) => break,
             Err(_) => break,
         }
@@ -235,7 +239,7 @@ async fn test_live_microphone() {
             }
             transcription = transcriptions.recv() => {
                 match transcription {
-                    Some(chunk) => {
+                    Some(Ok(chunk)) => {
                         transcriptions_received += 1;
                         if chunk.is_partial {
                             print!("\r[partial] {}", chunk.text);
@@ -245,6 +249,10 @@ async fn test_live_microphone() {
                             println!("\n[final] {}", chunk.text);
                         }
                     }
+                    Some(Err(e)) => {
+                        println!("[error] Transcription error: {}", e);
+                        break;
+                    }
                     None => {
                         println!("[debug] Transcription stream ended");
                         break;