SAY

Transforms text in a spreadsheet into AI-generated voiceover via OpenAI TTS or ElevenLabs voices

Perfect for generating text-to-speech using AI models from OpenAI and ElevenLabs, leaders in the voice-generating industry. This function enables batch generation of phrases within Google Sheets. The combination of AI voiceovers and spreadsheets makes it suitable for use in the gaming industry, education, language learning, and the creation of faceless TikTok or YouTube content.

It can generate text-to-speech in any of those languages: Chinese, Spanish, Hindi, Portuguese, French, German, Japanese, Arabic, Russian, Korean, Indonesian, Italian, Dutch, Turkish, Polish, Swedish, Filipino, Malay, Romanian, Ukrainian, Greek, Czech, Danish, Finnish, Bulgarian, Croatian, Slovak, Tamil.

Examples

Arguments List

=SAY_X(text, [voiceid], [options])
=SAY_Y(text, [voiceid], [options])

Options List

Particular object at the end of a formula which contains optional parameters values in the form {"parameter_name",value; "next_parameter_name",value}
For Example:{"speed",100; "stability",66; "similarity_boost",62; "style",5; "use_speaker_boost",TRUE; "optimize_streaming_latency",1}

speed

Speed of the generated speech, with 100 representing normal speed. Speeds from 50 to 99 are slowed down, while speeds from 101 to 150 are sped up.

For Example: 100, 110, or 90

stability

Adjusts the variability of ElevenLabs voices and does not apply to OpenAI TTS models.
The scale ranges from 100 (more variable) to 0 (more stable) for speech generation. Lower values may result in more expressive speech, with variations in output between regenerations, but could also introduce instabilities. Higher values ensure more consistency across regenerations but may render the voice somewhat monotone.

For Example: 60, 100, or 0

similarity

Fine-tunes the resemblance of ElevenLabs voices and is not relevant for OpenAI TTS models.
Low values should be used to minimize background artifacts in the generated speech. High settings enhance the voice's clarity and its similarity to the target speaker. However, very high values may introduce artifacts. It is advisable to experiment with this setting to determine the optimal value.

For Example: 70, 100, or 0

style

Fine-tunes the degree of style exaggeration in ElevenLabs voices; this setting does not apply to OpenAI TTS models.
High values are advisable for those seeking to accentuate the speech style. However, be aware that increasing values may introduce greater instability into the generated speech.

For Example: 0, 100, or 20

similarity_boost

When activated, makes the speech sound more like the original voice with ElevenLabs, but it might slow down how fast the speech is created. This setting doesn't work with OpenAI TTS models.

Options: TRUE or FALSE

optimize_streaming_latency

Latency optimizations at some cost to quality for the ElevenLabs model. These do not apply to OpenAI TTS
0 (default): no latency optimizations.
1: regular latency optimizations (~50% of possible latency improvement of option 3).
2: strong latency optimizations (~75% of possible latency improvement of option 3).
3: max latency optimizations.
4: max latency optimizations, but also with text normalizer turned off for even more latency savings (best latency, but can mispronounce, e.g., numbers and dates).

Options: 0, 1, 2, 3 or 4

voiceid2

A Voice ID from the extensive ElevenLabs library to use instead of the default voice.

List of voices: api.elevenlabs.io/v1/voices
User's voice library: elevenlabs.io/voice-lab

For Example: "zrHiDhphv9ZnVXBqCLjz", "zcAOhNBS3c14rBihAFp1", "z9fAnlkpzviPz146aGWa", or "yoZ06aMxZJJ28mfd3POQ"

model

Specifies the model used for generating text-to-speech voiceovers, including OpenAI TTS models and the latest model from ElevenLabs. By default, the model depends on the chosen voice.

Options: "default", "tts-1", "tts-1-hd", or "eleven_multilingual_v2"

format

Output format of the mp3 file with generated text-to-speech audio.
mp3_44100_64: 44.1kHz sample rate at 64kbps.
mp3_44100_96: 44.1kHz sample rate at 96kbps.
mp3_44100_128 (default): 44.1kHz sample rate at 128kbps.
mp3_44100_192: 44.1kHz sample rate at 192kbps.
Options: "mp3_44100_128", "mp3_44100_64", "mp3_44100_96", or "mp3_44100_192"

filename

Name of the output file you will receive, intended to clarify the generated voice-over MP3 file and to enhance the organization of the spreadsheet.

For Example: "Dave-Speech-1" or A2

generate

If TRUE, the formula executes the text-to-speech generation. If FALSE, a warning message is displayed. To avoid accidental costs, connect this parameter to a Google Sheets cell with a checkbox or a validating formula.

For Example: TRUE, FALSE, or N$8

generate_warning

When 'generate' is set to FALSE, this message is displayed. It's recommended to provide directions on how to enable voice generation using ElevenLabs or OpenAI TTS.
Add the [STOP] codeword to stop processing related formulas, which remains hidden from the user. Input "default" for a standard message.

For Example: "default", "Choose a voice", or B18