diff --git a/src/calibre/ebooks/oeb/polish/tts.py b/src/calibre/ebooks/oeb/polish/tts.py
index 3fa922c202..f3f32d4256 100644
--- a/src/calibre/ebooks/oeb/polish/tts.py
+++ b/src/calibre/ebooks/oeb/polish/tts.py
@@ -473,7 +473,7 @@ def embed_tts(container, report_progress=None, parent_widget=None):
         pos = 0
         wav = io.BytesIO()
         wav.write(wav_header_for_pcm_data(size_of_audio_data, HIGH_QUALITY_SAMPLE_RATE))
-        afitem = container.generate_item(name + '.mp4', id_prefix='tts-')
+        afitem = container.generate_item(name + '.m4a', id_prefix='tts-')
         pfd.audio_file_name = container.href_to_name(afitem.get('href'), container.opf_name)
         smilitem = container.generate_item(name + '.smil', id_prefix='smil-')
         pfd.smil_file_name = container.href_to_name(smilitem.get('href'), container.opf_name)
@@ -496,8 +496,8 @@ def embed_tts(container, report_progress=None, parent_widget=None):
             wav.write(audio_data)
             make_par(container, seq, html_href, audio_href, s.elem_id, pos, duration)
         wav.seek(0)
-        with container.open(pfd.audio_file_name, 'wb') as mp4:
-            transcode_single_audio_stream(wav, mp4)
+        with container.open(pfd.audio_file_name, 'wb') as m4a:
+            transcode_single_audio_stream(wav, m4a)
         container.pretty_print.add(pfd.smil_file_name)
         container.dirty(pfd.smil_file_name)
         container.serialize_item(pfd.smil_file_name)
diff --git a/src/calibre/gui2/tts/piper.py b/src/calibre/gui2/tts/piper.py
index b9a1e64b2d..8279897eab 100644
--- a/src/calibre/gui2/tts/piper.py
+++ b/src/calibre/gui2/tts/piper.py
@@ -780,10 +780,10 @@ def develop_embedded():
         sz += len(data)
     all_data[0] = wav_header_for_pcm_data(sz, HIGH_QUALITY_SAMPLE_RATE)
     wav = ReadOnlyFileBuffer(b''.join(all_data), name='tts.wav')
-    mp4 = io.BytesIO()
-    mp4.name = 'tts.mp4'
-    transcode_single_audio_stream(wav, mp4)
-    subprocess.run(['mpv', '-'], input=mp4.getvalue())
+    m4a = io.BytesIO()
+    m4a.name = 'tts.m4a'
+    transcode_single_audio_stream(wav, m4a)
+    subprocess.run(['mpv', '-'], input=m4a.getvalue())
 
 
 def develop():
diff --git a/src/calibre/utils/ffmpeg.c b/src/calibre/utils/ffmpeg.c
index 286f85fc67..dd1b73dbff 100644
--- a/src/calibre/utils/ffmpeg.c
+++ b/src/calibre/utils/ffmpeg.c
@@ -221,7 +221,7 @@ open_output_file(Transcoder *t) {
     if (!(t->ofmt_ctx->oformat = av_guess_format(t->container_format, t->output_filename, NULL))) {
         if (t->container_format[0] || t->output_filename[0]) return set_error(t, PyExc_KeyError,
             "Could not determine container format for output filename: %s and container format name: %s", t->output_filename, t->container_format);
-        if (!(t->ofmt_ctx->oformat = av_guess_format("mp4", "file.mp4", NULL))) return set_error(t, PyExc_Exception, "ffmpeg is missing support for the MP4 container format");
+        if (!(t->ofmt_ctx->oformat = av_guess_format("m4a", "file.m4a", NULL))) return set_error(t, PyExc_Exception, "ffmpeg is missing support for the m4a container format");
 
     }
     const AVCodec *output_codec = NULL;