gTTS (gtts.gTTS)
- class gtts.tts.gTTS(text, tld='com', lang='en', slow=False, lang_check=True, pre_processor_funcs=[<function tone_marks>, <function end_of_line>, <function abbreviations>, <function word_sub>], tokenizer_func=<bound method Tokenizer.run of re.compile('(?<=\\?).|(?<=!).|(?<=?).|(?<=!).|(?<!\\.[a-z])\\. |(?<!\\.[a-z]), |(?<!\\d):|—|‥|\\)|:|…|。|¿|¡|\\[|、|\\\n|,|،|;|\\]|\\(', re.IGNORECASE) from: [<function tone_marks>, <function period_comma>, <function colon>, <function other_punctuation>]>, timeout=None)[source]
gTTS – Google Text-to-Speech.
An interface to Google Translate’s Text-to-Speech API.
- Parameters:
text (string) – The text to be read.
tld (string) – Top-level domain for the Google Translate host,i.e https://translate.google.<tld>. Different Google domainscan produce different localized ‘accents’ for a givenlanguage. This is also useful when
google.com
might be blockedwithin a network but a local or different Google host(e.g.google.com.hk
) is not. Default iscom
.lang (string, optional) – The language (IETF language tag) toread the text in. Default is
en
.slow (bool, optional) – Reads text more slowly. Defaults to
False
.lang_check (bool, optional) – Strictly enforce an existing
lang
,to catch a language error early. If set toTrue
,aValueError
is raised iflang
doesn’t exist.Settinglang_check
toFalse
skips Web requests(to validate language) and therefore speeds up instantiation.Default isTrue
.pre_processor_funcs (list) –
A list of zero or more functions that arecalled to transform (pre-process) text before tokenizing. Thosefunctions must take a string and return a string. Defaults to:
[ pre_processors.tone_marks, pre_processors.end_of_line, pre_processors.abbreviations, pre_processors.word_sub]
tokenizer_func (callable) –
A function that takes in a string andreturns a list of string (tokens). Defaults to:
See AlsoText-to-Speech AI: Lifelike Speech Synthesis | Google CloudErste Schritte mit Google Text-to-Speech mit PythonPre-processing and tokenizing — gTTS documentationText to Speech in Python [With Code Examples] - CodefatherTokenizer([ tokenizer_cases.tone_marks, tokenizer_cases.period_comma, tokenizer_cases.colon, tokenizer_cases.other_punctuation]).run
timeout (float or tuple, optional) – Seconds to wait for the server tosend data before giving up, as a float, or a
(connect timeout,read timeout)
tuple.None
will wait forever (default).
See also
Pre-processing and tokenizing
- Raises:
AssertionError – When
text
isNone
or empty; when there’s nothing left to speak after pre-processing, tokenizing and cleaning.ValueError – When
lang_check
isTrue
andlang
is not supported.RuntimeError – When
lang_check
isTrue
but there’s an error loading the languages dictionary.
- get_bodies()[source]
Get TTS API request bodies(s) that would be sent to the TTS API.
- Returns:
A list of TTS API request bodies to make.
- Return type:
list
- save(savefile)[source]
Do the TTS API request and write result to file.
- Parameters:
savefile (string) – The path and file name to save the
mp3
to.- Raises:
gTTSError – When there’s an error with the API request.
- stream()[source]
Do the TTS API request(s) and stream bytes
- Raises:
gTTSError – When there’s an error with the API request.
- write_to_fp(fp)[source]
Do the TTS API request(s) and write bytes to a file-like object.
- Parameters:
fp (file object) – Any file-like object to write the
mp3
to.- Raises:
gTTSError – When there’s an error with the API request.
TypeError – When
fp
is not a file-like object that takes bytes.
- exception gtts.tts.gTTSError(msg=None, **kwargs)[source]
Exception that uses context to present a meaningful error message
- infer_msg(tts, rsp=None)[source]
Attempt to guess what went wrong by using knowninformation (e.g. http response) and observed behaviour
Languages (gtts.lang)
Note
The easiest way to get a list of available languages is to print themwith gtts-cli --all
- gtts.lang.tts_langs()[source]
Languages Google Text-to-Speech supports.
- Returns:
A dictionary of the type { ‘<lang>’: ‘<name>’}
Where <lang> is an IETF language tag such as en or zh-TW,and <name> is the full English name of the language, such asEnglish or Chinese (Mandarin/Taiwan).
- Return type:
dict
The dictionary returned combines languages from two origins:
Languages fetched from Google Translate (pre-generated in
gtts.langs
)Languages that are undocumented variations that were observed to work andpresent different dialects or accents.
Localized ‘accents’
For a given language, Google Translate text-to-speech can speak in differentlocal ‘accents’ depending on the Google domain (google.<tld>
) of the request,with some examples shown in the table below.
Note
This is an incomplete list. Try different combinations of language codes andknown localized Google domains. Feelfree to add new combinations to this list via a Pull Request!
Note
The default tld
is com
which will use the local language accent(according to Google), if available, based on your geographical network location.For example, lang="en"
with the default tld
will sound British English.
Local accent | Language code ( | Top-level domain ( |
---|---|---|
English (Australia) |
|
|
English (United Kingdom) |
|
|
English (United States) |
|
|
English (Canada) |
|
|
English (India) |
|
|
English (Ireland) |
|
|
English (South Africa) |
|
|
English (Nigeria) |
|
|
French (Canada) |
|
|
French (France) |
|
|
Mandarin (China Mainland) |
| any |
Mandarin (Taiwan) |
| any |
Portuguese (Brazil) |
|
|
Portuguese (Portugal) |
|
|
Spanish (Mexico) |
|
|
Spanish (Spain) |
|
|
Spanish (United States) |
|
|
Examples
Write ‘hello’ in English to hello.mp3
:
>>> from gtts import gTTS>>> tts = gTTS('hello', lang='en')>>> tts.save('hello.mp3')
Write ‘hello’ in Australian English to hello.mp3
:
>>> from gtts import gTTS>>> tts = gTTS('hello', lang='en', tld='com.au')>>> tts.save('hello.mp3')
Write ‘hello bonjour’ in English then French to hello_bonjour.mp3
:
>>> from gtts import gTTS>>> tts_en = gTTS('hello', lang='en')>>> tts_fr = gTTS('bonjour', lang='fr')>>>>>> with open('hello_bonjour.mp3', 'wb') as f:... tts_en.write_to_fp(f)... tts_fr.write_to_fp(f)
Playing sound directly
There’s quite a few libraries that do this. Write ‘hello’ to a file-like objectto do further manipulation::
>>> from gtts import gTTS>>> from io import BytesIO>>>>>> mp3_fp = BytesIO()>>> tts = gTTS('hello', lang='en')>>> tts.write_to_fp(mp3_fp)>>>>>> # Load `mp3_fp` as an mp3 file in>>> # the audio library of your choice
Note
See Issue #26 fora discussion and examples of direct playback using various methods.
Logging
gtts
does logging using the standard Python logging module. The following loggers are available:
gtts.tts
Logger used for the
gTTS
classgtts.lang
Logger used for the
lang
module (language fetching)gtts
Upstream logger for all of the above