feat(backend,frontend,mcp): any voice or round video can be transcribed on request, media is found by its bytes

frontend/src/lib/components/MessageMedia.svelte

@@ -167,7 +167,12 @@
           url={ready.url}
         />
       {:else if ready && kind === "video_note"}
-        <VideoNote transcript={ready.transcript} url={ready.url} />
+        <VideoNote
+          chatId={message.chat_id}
+          messageId={message.message_id}
+          transcript={ready.transcript}
+          url={ready.url}
+        />
       {:else if ready && kind === "audio"}
         <AudioFile {own} title={fileName} url={ready.url} />
       {:else if ready && isImage}
@@ -210,6 +215,27 @@
           <Icon name="document" size="1.25rem" />
           <span>{label}</span>
         </button>
+      {:else if media?.state === "not-downloaded" && kind === "voice"}
+        <VoiceMessage
+          chatId={message.chat_id}
+          length={ref?.duration}
+          messageId={message.message_id}
+          ondownload={queue}
+          {own}
+          transcript={media.transcript}
+          url={null}
+          {waiting}
+        />
+      {:else if media?.state === "not-downloaded" && kind === "video_note"}
+        <VideoNote
+          chatId={message.chat_id}
+          length={ref?.duration}
+          messageId={message.message_id}
+          ondownload={queue}
+          transcript={media.transcript}
+          url={null}
+          {waiting}
+        />
       {:else if media?.state === "not-downloaded" && vk !== "other"}
         <button class="media-placeholder" onclick={queue} type="button">
           <Icon name={waiting ? "timer" : "download"} size="1.5rem" />