{%- for message in messages %}
  {%- if not message['role'] is string or not message['role'].isdigit() %}
    {{- raise_exception("Role must be a stringified integer speaker id.") }}
  {%- endif %}
  {%- set content = message['content'] %}
  {%- if content is not iterable or content is string %}
    {{- raise_exception("Content must be a list.") }}
  {%- endif %}
  {%- set texts  = content | selectattr('type','equalto','text')  | list %}
  {%- set audios = content | selectattr('type','equalto','audio') | list %}

  {%- if loop.last %}
    {%- if texts|length != 1 or audios|length != 0 %}
      {{- raise_exception("Last message must have one text and no audio.") }}
    {%- endif %}
  {%- else %}
    {%- if texts|length != 1 or audios|length != 1 %}
      {{- raise_exception("Each non-last message must have one text and one audio.") }}
    {%- endif %}
  {%- endif %}
{%- endfor %}

{{- bos_token }}
{%- for message in messages %}
  {%- set text_item = message['content'] | selectattr('type','equalto','text') | list | first %}
  {%- if not loop.last -%}
    [{{ message['role'] }}]{{ text_item['text'] }}{{ " " }}
  {%- else -%}
    {{ text_item['text'] }}
  {%- endif -%}
{%- endfor %}
{{- eos_token }}

{%- for message in messages %}
  {%- if not loop.last %}
    {{- '<|AUDIO|>' }}
  {%- endif %}
{%- endfor %}