OpenAI's high-definition text-to-speech model, accessed via OrcaRouter's API at $30 per 1M tokens (zero markup).
openai/tts-1-hd is a text-to-speech model from OpenAI that converts text into high-definition natural speech. It is the enhanced version of the standard TTS-1 model, offering improved audio quality,…
OpenAI's TTS-1-HD supports multiple built-in voices, including alloy, echo, fable, onyx, nova, and shimmer. Each voice has a distinct tone, from deep and resonant to bright and energetic. The model also handles multiple languages, such as English, Spanish, French, German, Italian, Portuguese, and others, with natural accent and rhythm. You can specify the voice and language in the API request. The high-definition model improves accent accuracy and emotional range compared to the standard version. For a full list of supported languages, refer to OpenAI's official documentation.
openai/tts-1-hd is optimized for lower latency than the standard TTS-1, making it suitable for real-time applications like voice assistants and live captioning. However, the exact latency depends on factors such as text length, network speed, and server load. The model supports streaming responses via the API, allowing you to start playback before the entire audio is generated. This reduces perceived delay. For near-instantaneous responses, consider using the standard TTS-1 model, which prioritizes speed over quality. OrcaRouter's stable infrastructure ensures minimal added latency.
If your application does not require premium audio quality, consider using OpenAI's standard TTS-1 model or other budget-friendly TTS providers. For example, when generating speech for internal tests, low-fidelity voice alerts, or character-limited scenarios, a cheaper model reduces costs. Additionally, if you are experimenting with many variations or need extremely low latency, TTS-1 may be more suitable. openai/tts-1-hd is overkill for simple notifications. Evaluate your quality threshold and cost tolerance: the HD model costs the same per token as the standard version on OrcaRouter, but its output may incur higher token counts for longer audio.
OpenAI's TTS-1-HD delivers higher fidelity speech with improved clarity, more natural prosody, and reduced clicking or buzz artifacts. It better captures emotional tone and handles punctuation and pauses more accurately. While no exact benchmark scores are provided, user and developer reports indicate noticeably better subjective quality. The model is particularly effective for long-form content like audiobooks, where consistent voice quality matters. For simple short phrases, the difference may be subtle. The trade-off is slightly higher computational cost, but pricing is identical per token.
Despite its quality, openai/tts-1-hd has limitations. It may still produce occasional mispronunciations, especially for uncommon names, foreign words, or technical terms. The model cannot generate singing, sound effects, or non-speech audio. It also has a maximum text input length (context window) per request, though the exact limit is not publicly detailed; very long inputs may need to be split and concatenated. The model lacks support for custom voice cloning through the standard API. Additionally, it is not designed to control speaking rate or pitch with fine granularity. For such advanced features, consider dedicated speech synthesis platforms.
Speed depends on the text length and requested audio format. openai/tts-1-hd is optimized for lower latency than its predecessor but is not the fastest option available. For typical sentences, response times are sub-second under normal conditions. Streaming capability allows partial results. OrcaRouter's API itself imposes negligible overhead because it proxies requests directly to OpenAI. For real-time applications where every millisecond matters, the standard TTS-1 model (or non-HD version) may deliver faster first audio bytes. Consider benchmarking with your typical payload to determine acceptable performance.
openai/tts-1-hd is priced at $30.00 per 1 million input tokens and $30.00 per 1 million output tokens. The input tokens correspond to the text you submit, while output tokens represent the generated audio. This is the provider rate; OrcaRouter adds zero markup. You are billed only for actual usage. Tokens are counted on both input and output, but because audio output is inherently longer in duration, the output token cost may dominate. For example, a 1,000-character text might cost roughly $0.0012 in input tokens, while the output (say 30 seconds of audio) could cost more.
On OrcaRouter, both TTS-1 and TTS-1-HD have the same per-token price, so the cost difference is not in unit price but in token usage. Because TTS-1-HD may produce higher-quality audio with different compression settings, its output token count could be similar to TTS-1 for the same text. However, if you require fewer retries due to better quality, the HD model may be more cost-effective overall. Other TTS providers may offer lower per-character rates but with lower quality. Weigh the importance of audio fidelity against budget. For high-volume applications, even small percentage differences matter.
OrcaRouter does not provide caching for TTS responses because each request can have different voice, language, or text. However, it passes the provider's pricing without any additional fees. There are no tiered discounts or volume pricing through OrcaRouter for this model; costs scale linearly with usage. If you anticipate very high usage, you might consider contacting OpenAI directly for enterprise pricing, but through OrcaRouter you benefit from simple pay-as-you-go billing. No minimum commitments or hidden costs apply.
You can use any OpenAI-compatible client (e.g., Python openai library) by setting the base URL to https://api.orcarouter.ai/v1. Include the model ID "openai/tts-1-hd" in your request. For example, using Python: client = OpenAI(base_url='https://api.orcarouter.ai/v1', api_key='your_orcarouter_key'). Then call client.audio.speech.create(model='openai/tts-1-hd', voice='alloy', input='Hello world', response_format='mp3'). OrcaRouter forwards the request to OpenAI and returns the audio file. Ensure your API key is from OrcaRouter, not directly from OpenAI.
The API supports several parameters: model (required: openai/tts-1-hd), input (the text to speak), voice (one of alloy, echo, fable, onyx, nova, shimmer), response_format (mp3, opus, aac, flac), speed (a float from 0.25 to 4.0, default 1.0), and optional streaming boolean. You can also set language to improve pronunciation for certain languages. OrcaRouter passes these through unmodified. Parameters like temperature or top_p are not applicable for TTS models. Refer to OpenAI's audio API documentation for full details.
Yes. Migration requires only two changes: replace the base URL from 'https://api.openai.com/v1' to 'https://api.orcarouter.ai/v1', and use an OrcaRouter API key instead of an OpenAI key. The request and response structures are identical. There is no need to modify your code logic or parameter names. OrcaRouter also supports the same streaming and error handling conventions. This makes switching straightforward for developers who want to manage multiple models through a single gateway.
The primary difference is audio quality. TTS-1-HD is designed for higher fidelity with better clarity and more natural intonation, while TTS-1 is optimized for lower latency and faster generation. Both share the same pricing per token on OrcaRouter. The HD model consumes slightly more computational resources on OpenAI's backend, but the token count for a given input text is identical. For short, simple phrases, the quality difference may be negligible; for long-form or emotional content, the HD version is noticeably better. Choose based on your quality and speed requirements.
ElevenLabs offers highly expressive speech with voice cloning capabilities but at a higher per-character cost. Google Cloud TTS provides a broad range of voices and SSML support but may not match the naturalness of OpenAI's HD model. openai/tts-1-hd strikes a balance between quality and cost, with a straightforward token-based pricing model. It does not support custom voice cloning through the standard API, which ElevenLabs excels at. Google's model offers more language coverage and fine-grained control. Through OrcaRouter, you can compare costs directly by running tests with your typical text lengths.
Use openai/tts-1-hd when your application's success depends on audio quality—for example, if you are creating professional content, user-facing voice assistants where first impressions matter, or accessibility tools requiring clear speech. Avoid it if your users cannot distinguish quality differences, such as in internal notification systems or when the output will be heavily compressed. Also consider that on OrcaRouter the price per token is the same for TTS-1 and TTS-1-HD, so the HD model doesn't penalize you on unit cost; you only pay more if longer audio is generated due to higher quality settings.
OpenAI-compatible — keep the SDK you already use
https://api.orcarouter.ai/v1response_formatspeedvoice| Input / 1M tokens | $30.00 |
| Output / 1M tokens | $30.00 |
| Currency | USD |
Estimate based on list price
Estimate only — actual token counts depend on the provider's tokenizer.
GET /api/public/models/openai/tts-1-hdOpen @misc{orcarouter_tts_1_hd,
title = {openai/tts-1-hd API},
author = {openai},
year = {n.d.},
howpublished = {OrcaRouter},
url = {https://www.orcarouter.ai/models/openai/tts-1-hd}
}openai. (n.d.). openai/tts-1-hd API. OrcaRouter. https://www.orcarouter.ai/models/openai/tts-1-hd