Add response attribute to TextToSpeechClient class

[jpwroberts]

Feature Description

Ability to query the response headers, specifically to get the request-id.

The client.TextToSpeechClient class doesn't expose the HTTPS response object, nor the request-id. That makes it difficult to populate the previous_request_ids argument to methods such as convert.

Including the response as an attribute would allow generational conditioning from the Python API, as in the code below:

previous_request_ids = deque(maxlen=MAX_REQUEST_IDS)

# The following code is in a loop iterating over the lines of text.

try:
    previous_text = OMIT if is_first_paragraph or len(text) == 1 else ' '.join(text[:line_no])
    next_text = OMIT if is_last_paragraph or len(text) == 1 else ' '.join(text[line_no + 1:])

    previous_request_argument = previous_request_ids if ((len(previous_request_ids) > 0)
                                                         and condition_on_past_generations) else OMIT

    audio = client.text_to_speech.convert(
        voice_id=voice_id,
        optimize_streaming_latency=None,
        output_format=output_format,
        text=line,
        model_id=model,
        voice_settings=settings,
        previous_text=previous_text,
        next_text=next_text,
        previous_request_ids=previous_request_argument,
        pronunciation_dictionary_locators=pronunciation_dictionary_locators
    )
except Exception as e:
    log_exception(e)
    print(e)
    break
else:
    for chunk in audio:
        audio_segments.append(chunk)
    if condition_on_past_generations:
        request_id = client.text_to_speech.response.headers['request-id']    # <------------------
        if request_id:
            previous_request_ids.append(request_id)

Use Case

Passing previous request IDs to the various methods of the TextToSpeechClient object.

The methods accept a previous_request_ids argument, but the class doesn't provide an obvious way to obtain the request IDs.

Alternatives Considered

At the moment I edit client.py to add the attribute:

class TextToSpeechClient:
    def __init__(self, *, client_wrapper: SyncClientWrapper):
        self._client_wrapper = client_wrapper
        self.response = None

then update self.response from _response in each of the methods. I have to make that change again each time the elevenlabs package is updated.

Additional Context

No response

[PaulAsjes]

Excellent timing, we noticed this as well and are working on a fix right now. Hope to have something out later this week.

[jpwroberts]

Thanks :-). The change is quite trivial, but as I generate Windows and Linux versions of my program I have to make the change twice each time.