Speaking issues Coqui TTS Tacotron2 DDC model: Difference between revisions

Latest revision as of 23:44, 12 November 2021

Most models are trained with a dot, exclamation or question mark at the end. So always end a sentence to avoid model synthesizing weird output.

"ah" at end of sentence generally produces strange results. Short names produce a 12 second clip.

Examples:

If at the end of the input, adding punctuation to the end synthesizes correctly:

Example "Nelson Mandela" > "Nelson Mandela."

To speak acronyms as letters it needs to be formatted as:

"A. B. C. news"

Not:

"ABC news" "A.B.C. news"

@@ Line 2: / Line 2: @@
 [[Category:TTS models]]
 [[Category:Pronunciation issues]]
+== General ==
+Most models are trained with a dot, exclamation or question mark at the end. So always end a sentence to avoid model synthesizing weird output.
 *TTS model: tts_models/en/ljspeech/tacotron2-DDC
 *Vocoder model: vocoder_models/en/ljspeech/hifigan_v2
-== General ==
+==Input string formatting==
-Most models are trained with a dot, exclamation or question mark at the end. So always end a sentence to avoid model synthesizing weird output.
+===Phrases ending in "ah"===
+"ah" at end of sentence generally produces strange results. Short names produce a 12 second clip.
+Examples:
+* Nelson Mandela
+* pergola
+==== Mitigation ====
+If at the end of the input, adding punctuation to the end synthesizes correctly:
+Example "Nelson Mandela" > "Nelson Mandela."
+=== Acronyms ===
+To speak acronyms as letters it needs to be formatted as:<blockquote>"A. B. C. news"</blockquote>Not:<blockquote>"ABC news"
-==nelson mandela==
+"A.B.C. news"</blockquote>
-"ah" at end of sentence generally produces strange results. Just the name produces a 12 second clip.
-== pergola ==
+== Mispronounced Words ==
-"ah" at end of sentence generally produces strange results. Just the name produces a 12 second clip.
-== ABC news ==
+* video
-To speak acronyms as letters it needs to be formatted as "A. B. C. news"