release myna

Replace the Myna-Hokkien demo responses for samples 01, 07, 08, and 10 and align the documented built-in audio prompt with the release default.

README.md

@@ -116,7 +116,7 @@ audio_only = model.generate(text="講一句歡迎詞。", return_text=False, ret
 If `prompt=` is omitted for audio input, MynaHokkien uses this built-in prompt:
 
 ```text
-Answer or respond to what is said in the audio, in spoken Singapore Hokkien. Do NOT repeat or transcribe it.
+Listen to the spoken Hokkien and reply naturally in concise Singapore Hokkien. Always answer in colloquial Singapore Hokkien written in Hanji. Never answer in Mandarin or English. Do not repeat or transcribe the input; respond to it directly.
 ```
 
 To override it, pass a different instruction through `prompt=`:

demo_samples/release_csv/01/01_mynahokkien.wav

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:aca3b47f879c3c8037122a1bb70679570b3eb0647921a459faf137fd65a1429f
-size 256214
+oid sha256:a57503d25dc87397bd02e5ee4d11d344038433513663360b64897dd35c69bc2a
+size 240854

demo_samples/release_csv/07/07_mynahokkien.wav

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:e7859a64e09ff1bc96909669b580f474693244ada191bd2f03f7fb6d888e5b93
-size 171734
+oid sha256:212ccb076a80d156ae80a3051c89ef5f0ffa09aec03441159589d2e50a236a6b
+size 237014

demo_samples/release_csv/08/08_mynahokkien.wav

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:cc8239ad2521118268bb610680d306453fd4118d04232e3423683c23a29065cd
-size 202454
+oid sha256:f0c599af7810425eccf80951f71746d741ebf6c775dabd012d800d292c997fa6
+size 233174

demo_samples/release_csv/10/10_mynahokkien.wav

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:f62e16b58aed75a7a42dc586633f11e668d68d1a48d0783df5981305167eb9ee
-size 217814
+oid sha256:ad170c087d112e0c01f2c61e0fd39c12661b7615a16c079bbcdab5103e576cf6
+size 221654