Speechify Pricing Alternative for Language Learning Teams

By VerbaLingua editorial team · Updated 2026-11-18

Language learning teams need reliable synthetic speech that scales with curriculum growth without unpredictable costs. VerbaLingua offers a transparent, character-based pricing model designed for product teams building multilingual applications. Unlike subscription-heavy alternatives, our API charges only for what you generate, making expenses predictable and easy to forecast.

This page outlines how to integrate our text-to-speech API into your learning platform. We cover voice selection, language support, technical limits, and data privacy specifics relevant to education. Use this guide to evaluate whether our straightforward API fits your team’s workflow before committing to a key.

Speechify Pricing Alternative for Language Learning Teams

Frequently asked questions

Is this the official Speechify API?

No. VerbaLingua is an independent storefront for its own text-to-speech API. It is not affiliated with Speechify. The endpoint structure matches the OpenAI speech endpoint, but the underlying engine, voices, and pricing are specific to VerbaLingua. You use VerbaLingua’s API key, not Speechify’s, to generate audio.

How does Speechify pricing compare to character-based APIs?

VerbaLingua charges $10 per 1 million characters, or $0.01 per 1,000 characters. This flat rate applies to both tts-1 and tts-1-hd models. Failed requests are free. You pay only for what you generate, with no monthly plans or minimum spend requirements.

Can I use these voices for commercial language learning apps?

Yes. The voices are based on licensed recordings from LibriTTS-R (CC BY 4.0) and public-domain audiobook readings. You can integrate them into commercial applications. Each generated file carries an inaudible watermark to mark it as synthetic speech, but there are no restrictions on commercial usage.

Does the API support SSML for pronunciation control?

No. The API does not support SSML. It also lacks phoneme dictionaries, custom pronunciation lexicons, or emotion tags. You control audio output using the voice preset, speed parameter (0.25–4.0), and optional language ISO code. For precise control, you must split text into smaller segments.

How many characters can I send in one request?

You can send up to 4096 characters per request. For longer texts, split them into parts of this size and join the resulting audio files. Spaces and punctuation count toward the character limit. The API processes requests sequentially, so queue times may apply during high load.

Are the voices cloned from real people?

No. The voices are presets based on licensed recordings from LibriTTS-R and LibriVox public-domain audiobook readings. We do not offer custom voice cloning for customers, nor do you upload your own voice. The voices are synthetic interpretations.

What happens if a request fails during synthesis?

If synthesis fails, the charge is returned in full. You are not billed for failed requests. This applies to any error returned by the API. You can retry the request or adjust parameters like text length or voice selection without incurring additional costs for the failed attempt.

Do you offer volume discounts for large language schools?

No. There are no volume discounts, monthly plans, or tiered pricing. The rate is a flat $10 per 1 million characters regardless of usage volume. There are no seats, minimum spends, or enterprise contracts. Pricing remains consistent whether you generate 1,000 or 10 million characters.

Can I stream audio while it generates?

No. The API does not support streaming, WebSockets, or real-time partial audio. You receive the complete audio file after synthesis finishes. Typical speed is roughly real time for long texts, but short phrases may take a few seconds. Requests are processed one after another.

Is there a free trial available?

Yes. Signing in with Google provides $0.50 of credit, valid for 7 days, allowing about 50,000 characters. Accounts created with email and password start without trial credit. You can top up with USDT on the TRON network to continue usage after the trial or credit expires.

How do I switch my existing app to this API?

Update your base_url to https://api.play99ab.com/v1 in the OpenAI SDK. Use your VerbaLingua API key (starting with tts-...) for authentication. The request shape matches the OpenAI speech endpoint. Ensure your code handles the 4096-character limit by splitting longer texts before sending them to the API.

Do you support Vietnamese or Thai for language learning?

No. Vietnamese, Thai, Indonesian, Ukrainian, Romanian, Czech, and Bengali are not supported. The API supports 23 languages, including English, Spanish, Chinese, and Japanese. If you need those specific languages, this API is not suitable. For non-English Latin-script text, set the language parameter.

What audio formats are supported?

The API supports mp3, wav, opus, flac, and pcm formats. Specify the desired format using the response_format field in your request. All generated files are returned directly in the response. There is no storage of generated audio; you must save it yourself if needed for your app.

Is my input text used to train your model?

No. Input text is used only to produce the audio and is not used for training. Generated audio is returned in the response and not stored. Request metadata, such as time, characters, and status, is kept for billing purposes. Your data remains private and is not part of the training set.

First request

curl

curl https://api.play99ab.com/v1/audio/speech \
  -H "Authorization: Bearer $TTS_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{"model":"tts-1","voice":"aria","input":"Hello from the API.","response_format":"mp3"}' \
  --output hello.mp3

Why switch from Speechify for language learning

Sound wave illustration for: Why switch from Speechify for language learning

Switching from Speechify to VerbaLingua helps you manage your language learning app’s audio budget. Our pricing is strictly character-based at $10 per million characters, ensuring you pay only for what you generate. This model eliminates the surprise fees often associated with premium voice tiers or usage spikes.

Transparent character-based pricing

Transparent character-based pricing

We charge $0.01 per 1,000 characters, including spaces and punctuation. This predictable rate applies to all 8 voice presets without tiered pricing. You know exactly what a 10,000-character lesson costs before you send the request.

  • Flat rate: $10 per 1 million characters
  • Includes spaces and punctuation in count
  • Same price for all voice models

No hidden voice licensing fees

All 8 presets are included in the base rate. All voices are based on licensed recordings from LibriTTS-R and LibriVox, so you can use them commercially without additional licensing checks.

Predictable monthly costs

With no monthly plans or minimum spend, your costs scale directly with user activity. Failed requests are free, and balances do not expire. This makes it easy to forecast monthly expenses for your language learning platform based on expected lesson volumes.

Core API capabilities

Sound wave illustration for: Core API capabilities

The API accepts standard JSON payloads via a POST request, mirroring the OpenAI speech endpoint structure. You send text, select a model and voice, and receive an audio file. This consistency allows teams to swap providers without rewriting their integration logic.

Simple JSON request format

Send a JSON body with fields like model, input, voice, and response_format. The endpoint handles the synthesis and returns the audio directly. Keep inputs under 4096 characters to ensure reliable processing.

OpenAI SDK compatibility

Use the official OpenAI SDKs by simply changing the base_url to https://api.play99ab.com/v1. Pass your API key in the Authorization header. This allows you to leverage existing client libraries without building a custom HTTP client from scratch.

Standard audio formats

Choose between MP3, WAV, Opus, FLAC, or PCM for your output. This flexibility supports various playback contexts, from mobile apps to web browsers, ensuring compatibility with your existing media pipelines.

Voice selection for learners

Sound wave illustration for: Voice selection for learners

Choosing the right voice is critical for language learning, where clarity and consistency directly impact learner comprehension. VerbaLingua provides eight distinct presets, each engineered for specific vocal qualities like warm narration or upbeat energy, ensuring your app maintains a consistent auditory identity across all modules.

These voices are not generic; they are based on licensed recordings from LibriTTS-R and LibriVox public-domain audiobooks. This foundation provides natural intonation for listening exercises.

8 distinct voice presets

We offer eight specific voice profiles to match different learning contexts. Female voices include Aria (warm, clear narrator), Nova (bright, upbeat), Luna (soft, storytelling), and Sage (calm, instructional). Male options feature Felix (energetic, promo), Milo (friendly, conversational), Atlas (neutral, news), and Orion (deep, documentary).

Each preset is optimized for specific tones, allowing you to assign voices based on lesson type. For example, use Aria for general reading exercises or Orion for documentary-style content. This variety helps keep learners engaged without breaking immersion.

  • Aria: Warm, clear narrator
  • Nova: Bright, upbeat
  • Luna: Soft, storytelling
  • Sage: Calm, instructional
  • Felix: Energetic, promo
  • Milo: Friendly, conversational
  • Atlas: Neutral, news
  • Orion: Deep, documentary

Gender-neutral options

While our presets are categorized by gender, every voice can speak any of our 23 supported languages. This flexibility allows you to select voices based on tone rather than strict gender binaries, giving your product team full control over auditory branding.

You are not locked into traditional associations. A neutral voice like Atlas works well for news-style language drills, while Milo’s conversational tone suits interactive dialogue practice, regardless of the learner’s demographic.

Clear pronunciation models

Our voices are built on licensed recordings from LibriTTS-R (CC BY 4.0) and LibriVox public-domain audiobooks. This source material ensures high-quality, natural intonation and clear pronunciation across all supported languages.

If the language is not specified, the system guesses from the script (Latin defaults to English). For non-English Latin-script text, always set the language parameter to ensure accurate pronunciation, avoiding default English inflection on words like Spanish or German.

Language coverage

Sound wave illustration for: Language coverage

VerbaLingua supports 23 languages, covering major markets for language learning apps. Any voice can speak any supported language, allowing your learners to hear consistent character tones across Spanish, French, German, Japanese, and more. This flexibility simplifies voice management when scaling to new locales.

If you do not specify the language, the system guesses it from the script. Latin script defaults to English, so you must set the language field for other Latin-based languages like Dutch or Swedish. Vietnamese, Indonesian, Thai, Ukrainian, Romanian, Czech, Bengali, and others are not supported.

23 supported languages

Our API handles Arabic, Chinese, Danish, Dutch, English, Finnish, French, German, Greek, Hebrew, Hindi, Italian, Japanese, Korean, Malay, Norwegian, Polish, Portuguese, Russian, Spanish, Swahili, Swedish, and Turkish. Each language benefits from the same eight voice presets, ensuring a unified user experience across your entire application.

This coverage allows product teams to prototype multilingual features. You can test voice consistency without managing separate voice models for each locale, reducing technical debt in your language learning stack.

  • Arabic, Chinese, Danish, Dutch, English, Finnish, French, German, Greek, Hebrew, Hindi, Italian, Japanese, Korean, Malay, Norwegian, Polish, Portuguese, Russian, Spanish, Swahili, Swedish, Turkish

Non-English Latin script handling

When the language field is omitted, the API infers the language from the text script. Latin text defaults to English, which may cause mispronunciation for words in other Latin-script languages. Always explicitly set the language parameter for accurate pronunciation in French, German, or Portuguese.

This explicit control prevents subtle errors in educational content. Accurate pronunciation is critical for language learners who rely on audio to build correct phonetic associations. Setting the language ensures the voice engine applies the correct phoneme rules.

Dialect support limits

The API does not distinguish between regional dialects within a supported language. For example, Spanish voices will use a standard neutral accent rather than specific regional variations like Mexican or Argentine Spanish. Similarly, Portuguese does not differentiate between Brazilian and European variants.

This limitation is consistent across all 23 languages. If your language learning product requires specific dialects, you may need to curate text or accept a neutral standard. This approach keeps the API simple and predictable for integration.

Technical limits to know

Sound wave illustration for: Technical limits to know

Understanding the technical boundaries of the TTS API is essential for integrating text-to-speech into language learning applications without unexpected errors or latency spikes during student interactions.

4096 character input limit

Each API request accepts a maximum of 4096 characters for the input field. This limit applies to the raw text string, including spaces and punctuation. For longer lessons or dialogues, your application must split the text into chunks and concatenate the resulting audio files.

60 requests per minute

The API enforces a rate limit of 60 requests per minute per API key. New accounts start with a lower limit of 10 requests per minute until the first top-up is processed. If you exceed this threshold, requests may be throttled or rejected.

Real-time generation speed

Synthesis typically runs at roughly real-time speed, generating about 25 characters of text per second for longer inputs. Short phrases may take a few seconds to process. Requests are handled sequentially, so high concurrency can introduce queuing delays.

Technical limits to know

Audio quality and formats

Sound wave illustration for: Audio quality and formats

VerbaLingua delivers clean audio files via standard formats like MP3, WAV, Opus, FLAC, and PCM. Each request returns a downloadable file ready for immediate integration into your language learning app’s playback engine.

MP3 and WAV output

Choose the format that best fits your app’s storage and bandwidth constraints. MP3 offers efficient compression, while WAV provides lossless quality. Both formats are generated instantly upon successful synthesis.

  • MP3 for balanced quality and file size
  • WAV for uncompressed, high-fidelity playback
  • Opus, FLAC, and PCM also available

Synthetic watermarking

Every generated audio file contains an inaudible watermark that marks it as synthetic speech. This helps listeners distinguish AI-generated pronunciation guides from human recordings in your course materials.

No SSML tags

We do not support SSML (Speech Synthesis Markup Language) or custom pronunciation lexicons. Input text is processed as plain strings, so you cannot inject pause tags or phoneme overrides directly into the audio stream.

Developer experience

Sound wave illustration for: Developer experience

Integrating VerbaLingua into your language learning app is straightforward, leveraging familiar patterns if you are already using OpenAI’s infrastructure. You can test the endpoint immediately with a simple cURL command or migrate your existing client code with minimal adjustments.

The API accepts standard JSON bodies, allowing you to generate speech by sending text, voice selection, and format preferences. This consistency reduces boilerplate code and speeds up initial deployment for product teams focused on multilingual content delivery.

Quick start with cURL

You can verify your key and generate audio instantly using a single terminal command. This method requires no library installation, making it ideal for quick validation of voice models or debugging input text before integrating into your application layer.

  • Use POST https://api.play99ab.com/v1/audio/speech
  • Include Authorization: Bearer tts-... header
  • Send model, input, voice, and response_format in JSON

OpenAI SDK integration

Since the request shape matches the OpenAI speech endpoint, you can use official OpenAI SDKs by simply changing the base_url to https://api.play99ab.com/v1. This allows you to maintain your current code structure while switching providers without rewriting your audio generation logic.

Authentication is handled via a Bearer token in the header. Ensure your SDK configuration points to the correct base URL to avoid routing errors when switching between environments or providers.

Error handling basics

The API returns standard HTTP status codes to indicate success or failure. If synthesis fails, you receive a clear error message, and the request is free, ensuring you are not charged for unsuccessful generations.

Always check the response status before parsing the audio file. Implement retry logic for rate limit errors (429), as you are limited to 60 requests per minute per key, with a lower initial limit of 10 requests per minute.

Data privacy for education

Sound wave illustration for: Data privacy for education

VerbaLingua prioritizes student data protection by ensuring input text is used solely for synthesis and never for model training. This guarantees that unique educational content remains private and distinct from our dataset improvements.

No training on input text

All text sent to the API is processed exclusively to generate the requested audio file. We do not retain or analyze this input to improve our core engine, ensuring your lesson materials remain yours alone.

No audio storage

Generated audio files are returned directly in the API response and are not stored on our servers after delivery. Your students’ voiceovers are transient, reducing your long-term data liability and storage costs.

Minimal metadata retention

We keep only the necessary metadata—timestamp, character count, and status—strictly for billing purposes. No personal identifiers or content logs are retained beyond what is required to verify your usage and charges.

Get started today

Sound wave illustration for: Get started today

Create your account at /get-api-key/ using your Google identity to receive $0.50 in trial credit immediately. This provides approximately 50,000 characters of synthetic speech, valid for seven days, allowing you to test voice quality and integration speed without financial commitment. No credit card is required to begin.

For sustained usage, top up your balance using USDT on the TRON network (TRC-20). Payments of $5 or more are credited automatically within minutes. There are no monthly subscriptions, invoices, or VAT charges, keeping your cost structure simple and predictable as your language learning application scales.

Google sign-up bonus

Sign up with Google on the /get-api-key/ page to instantly receive $0.50 in credit. This allows you to generate roughly 50,000 characters of audio over seven days to validate the API’s performance for your specific use case before committing funds.

USDT top-up method

Fund your account directly with USDT on the TRON network (TRC-20). A minimum top-up of $5 triggers automatic crediting within minutes. This method supports unlimited usage without the friction of recurring billing cycles or manual invoice processing.

Balances never expire, and you only pay for the characters you actually consume at $0.01 per 1,000 characters. Failed requests are refunded in full, ensuring you never pay for unusable audio.

  • Supported: USDT (TRC-20) on the TRON network
  • Minimum top-up: $5
  • Processing time: Automatic within minutes

No credit card required

You can start generating speech immediately without linking a bank card or credit card. This reduces the barrier to entry for student developers and small language learning teams who want to test the API’s compatibility with their existing tech stack.

Your generated audio is returned in the response and not stored, while only minimal metadata is kept for billing. This approach ensures your learners’ data remains private while you manage your budget transparently.

Listen to the 8 voices

Each sample below was generated by this API from the same sentence. Users need intuitive, adaptive lessons that fit seamlessly into their daily routines for effective language acquisition.

Avatar of the voice aria
ariafemale
Avatar of the voice nova
novafemale
Avatar of the voice luna
lunafemale
Avatar of the voice sage
sagefemale
Avatar of the voice felix
felixmale
Avatar of the voice milo
milomale
Avatar of the voice atlas
atlasmale
Avatar of the voice orion
orionmale

23 languages

Any voice speaks any of these languages.

Sources and further reading

  1. Wikipedia
  2. openai-node on GitHub
  3. openai-python on GitHub
  4. TRON developer hub: TRC-20
  5. W3C WAI: making audio and video accessible
  6. curl documentation
  7. Python documentation
  8. LibriTTS-R paper (arXiv 2305.18802)
  9. MDN: Web Speech API
  10. Creative Commons Attribution 4.0
  11. LibriVox public-domain audiobooks
  12. OpenAI API reference: create speech

More on this site