Eval bug: mtmd prompt format is missing audio begin tag for Voxtral

[steampunque]

Name and Version

master

Operating systems

Linux

GGML backends

CUDA

Hardware

NA

Models

Voxtral

Problem description & steps to reproduce

Tracing flaky behavior of Voxtral for audio processing led to discovery that the [BEGIN_AUDIO] tag is not being used in the mtmd prompt template. It sort of works without it but is quite flaky. Recommend fix:

--- mtmd.cpp	2025-12-08 13:13:44.202285955 -0500
+++ mtmd.cpp.new	2025-12-08 13:13:29.850285270 -0500
@@ -330,10 +330,10 @@
             aud_beg = "<|audio_bos|>";
             aud_end = "<|audio_eos|>";
 
-        } else if (proj == PROJECTOR_TYPE_ULTRAVOX) {
+	} else if ((proj == PROJECTOR_TYPE_ULTRAVOX) ||
+		   (proj == PROJECTOR_TYPE_VOXTRAL)) {
             // [BEGIN_AUDIO] ... (embeddings) ...
             aud_beg = "[BEGIN_AUDIO]";
-
         }
     }

First Bad Commit

NA

Relevant log output

NA

[github-actions]

This issue was closed because it has been inactive for 14 days since being marked as stale.